The Best of Ask David: Volume 1
by David Luciani
published December 21, 2003

As we continue to bring new features to our new-look site, we went through all of our old issues of "Ask David" to find some of the questions that best gave insight into the kind of forecasting I do or approach I take to either thinking about baseball or fantasy baseball.  I didn't pick questions and answers that necessarily made us look good (or bad), though there are some of both here, but instead went for explanations where examples helped us to illustrate the point we were making, for better or worse.  I am always the first to admit that I will miss on some players but at the same time, I am absolutely not afraid to take risks because our average margin of error proves that, on average, we're doing our job well, even if there is the certain player who we will get completely wrong.

In particular, I want to encourage readers to check out two topics included here, one a question I answered about top prospects lists in March 2003 and the other, an answer about closer valuation on fantasy forms in February 2003.

I'm hoping readers will look at this set of some of the best questions and/or answers to get insight into my way of thinking as opposed to just glancing at them to see what came true and what didn't.  During this holiday season, for those of you who celebrate Christmas, I want to wish Merry Christmas and to all of you, the happiest of holidays and a fantastic new year!

From AskDL February 15, 2001:

Q. Why do you think Jeff Conine is going to get so much playing time and be so good all of a sudden?

A. First, in terms of the playing time, we actually have Conine playing about 60 games at third, about 20-25 games at each of first base and the outfield and another 30 or 40 games at DH, which works out to him getting almost a full season's worth of playing time despite the lack of a defined role in the lineup. We have to do this sort of forecasting, even though the reader doesn't see it, in order to make good forecasts for the errors category.

As for the performance, this is as tough to explain in a short essay as his projection was to build. Basically, we're looking at Conine's improved ability to make contact the past two years. His swing has not been shortened dramatically yet his ability to hit singles and doubles in prior years has not shown up in 1999 and 2000 as much as we believe it should have.

Readers afraid of actually seeing the math of the situation should skip beyond this question. For those who like this sort of detail, here's a more scientific explanation. Though our projection changes daily as we move towards the next publishing date, as of February 15th, we had Conine hitting 104 singles, 37 doubles, 4 triples and 23 home runs in 511 at bats. In those at bats, we had him striking out 55 times which means that essentially 456 times (511 - 55 = 456) he puts the ball in play. So, he singles 23% of the time he puts the ball in play, doubles 8% of the time, triples almost 1% of the time and homers about 5% of the time.

If we even look at Conine's career numbers, which is not a good basis for building a projection but a half-decent start, he has singled 24% of the time he has put the ball in play, doubled about 7% of the time, tripled about 1% of the time and homered about 4% of the time.

Conine's projection therefore represents an example of the sort of hidden statistics to which I often refer. When you look at things on a per at bat basis, you think that when we project Conine to hit .329 that we're projecting him to have a great year. In fact, we're projecting him to maintain his improved rate of making contact from 1999 and 2000. Beyond that, we actually have him singling at below his career average and doubling and homering at a slightly improved mark. We're just looking at things a bit differently than normal statistics analysis would allow.

Conine had an excellent year and this turned out to be one of our better forecasts, ironic because it was the first pre-season question we answered in 2001 in an "Ask David" issue.  Conine hit .311 with 14 home runs and 97 RBI (our final pre-season projection was for 82) so we did overestimate his power somewhat.  In terms of the playing time, he played 139 games and he assembled that many games doing almost exactly what we said, playing here and there and never really settling on one position in the Orioles' lineup.

Q. Every time a pitcher gets traded to the Rockies, his value goes down (i.e. Mike Hampton). This also should work in reverse (witness Kile last year).  I am looking at Julian Tavarez as a real sleeper.  He pitched surprising well last year as a starter, outperforming his 3 year averages. He may not be going to a pitcher's paradise in Wrigley, but it is not Coors. His grounder-to-fly ratio is exceptional which is a definite requirement for Wrigley.  Your projections show ERA and WHIP higher than last year instead of lower, as I would expect.  Am I missing something or are you?

A. Actually, neither. You're astute to recognize that a pitcher leaving Coors should experience an improvement but the improvement should be relative to your expectations or the upcoming season - not what he did the previous year.  Obviously, we made an adjustment to our forecasts the moment Tavarez left Colorado and had he returned to Coors, our forecast would be worse than it is now.

But there's more to making a forecast than looking at what a player did in the most recent season. We agree with you that Tavarez enjoyed an excellent season in 2000 when one considers where he pitched half of his games. But we also believe that it was not fully representative of his ability for a number of reasons.

Firstly, his strikeout to walk ratio was poor (62 strikeouts, 53 walks) and though Coors does have some effect on these numbers and he walked nine batters intentionally, the ratio doesn't exactly reveal that he's suddenly developed the tools to be an outstanding pitcher, which his 2000 ERA would imply given where he pitched. His ERA and W-L record was so good last year because he didn't allow many hits and though that area improved over prior seasons, I'm one who believes that he can't maintain the same sort of pace. I also think his control will be a problem and beyond all that, he's likely to be in the Cubs' rotation rather than in the bullpen. Only 12 of his 51 appearances last year were out of the Rockies' rotation and, of course, a reliever averages a better ERA than a starter because he doesn't get charged with mid-inning inherited runners that score. With all that considered, our forecast as of February 14th was that Tavarez would go 10-12 for the Cubs with a 4.98 ERA.  In this era and given his team, that is actually not that bad a season if he ends up having it.

Tavarez's ERA ended up at 4.52, higher than the 4.43 of 2000 and close to our (unpopular) final pre-season projection of 4.67, which was lower than the 4.98 that was being published at this time.  Tavarez went 10-9.

From AskDL April 16, 2001

Q, I have Eric Young, and a lot of people are asking me to trade him. Why do you only project 26 steals for him?  He is already more than 1/5 of the way there, and seems to be running often.

A. Early in the year, I often get these sort of questions and it's important to answer at least one in public.

Our latest forecast has Young stealing 26 bases but it's important to note that when our forecasts are published, this refers to the remainder of the 2001 season. This is not to say that we forecast him to finish with 26 stolen bases. This means that in addition to what he already had on the day we published the projection, we expected him to steal 26 more bases between that day and the end of the season. On that day, whether Young had 5 stolen bases or none, it would not be included in the 26 steals. If Young had 5 as of the day, when the latest issue when to press, then that would mean we projected him to finish with 31 stolen bases (5 in the books + the 26 remaining steals).

As the season progresses, all of our projected totals will shrink to gradually reduced amounts that reflect fewer games left in the books. So, if you see us projecting a 3.00 ERA pitcher who has started with four disaster outings, those disaster outings aren't included in the projected 3.00 ERA. That's important to know and it ensures that our forecasts remain useful throughout the season. What a player has already done is of no use, especially to fantasy players. You need to know what a player will do after you trade for him or insert him into your lineup. That's why we do things the way we do and I hope this clarifies things.

Eric Young ended up with 31 steals.

Q. You have been projecting A-Rod to hit in the high .300s and he is hitting .240 as I write this note and he is killing me on my fantasy roster. Despite your ranking sheets, I've benched him for now. At one point will you admit you're wrong?

A. Well, since you wrote me, A-Rod had a 6 RBI game and had pushed his average over .340 at one point.  If you benched him, then you missed that nice performance. Even if his average falls back the other direction, if he were hitting .240 at the end of May, we could still be in a position to rightfully say he'll hit .340 from June to September. It usually takes about 100-200 at bats for a forecast to dramatically change but we've done it quicker before. A-Rod is A-Rod and there's nothing to worry about yet.

If you give up on established players too early, that's a guaranteed way to lose in your fantasy league. In fact, I've known far too many fantasy GMs who have five or ten players on their roster start off with an ERA of 9.00 or a sub-.200 batting average. They bench or trade them for the hot starters and then wonder at season's end why they finished in the middle of the pack.

If you're relatively certain that a player's skill isn't where it once was, then by all means don't wait too long to get rid of him. But Alex Rodriguez could go 0 for April without proving that he's any bit different than he was with the Mariners. Obviously, his salary gives him absolute job security and his track record says that the performance will be there.

The worse part about situations such as yours is that if you benched A-Rod and now he gets hot, you don't get the great week. Then you get excited and activate him and he goes 2 for 20 the subsequent week. By the time you finish with the trends, you end up with about 300 at bats from him, a .240 average and a handful of home runs. It's no good to your fantasy chances to wait for players to have good weeks before activating them. You need to have the best players on your roster and let them do what they do. If you lose at year end, it will be because too many of your players had bad years, not bad weeks.

Rodriguez finished with a .318 average with 52 home runs and 135 RBI.

From AskDL May 6, 2001:

Q. With Luis Gonzalez having so many home runs, are you likely to make a change to his projection and if so, at what point in a season do you typically acknowledge that a player has changed abilities?

A. This is such a frequent and good question that we are going to make most of this issue focused on answering this question at length.  In the past, we’ve tried to give brief answers to questions directed to this column but this is an opportunity to properly respond to the question that, by far, is the most asked of any we receive during a season.

Luis Gonzalez, the player you mentioned in your question, is an excellent example that we can use for multiple reasons.  First, his projection is on the verge of being radically changed in the upcoming projection revisions this Sunday.  Indeed, the first issue in May is always historically the issue in which we see the most dramatic changes because we usually cross the threshold of accepting new data for players at the end of April.

Let us begin by setting aside all considerations of the effects of park, age, team and league.  In other words, for purposes of demonstrating a point, we will allow ourselves to pretend that the 2001 season is being played under the same conditions as previous seasons (it isn’t) and that Luis Gonzalez’s historical data in his career does not need to be modified for where he played, how old he was or what his role was.  Indeed, we actually do make these modifications when making forecasts.

Most readers familiar with my work also know that I look at power a bit differently than we normally see it.  I mentioned this in a previous issue but at a basic level, I look at the number of home runs a player hits per time making contact (which can be roughly estimated by using at bats and subtracting strikeouts).

Now, let’s look at Gonzalez’s power ratings, as explained above, throughout his career:

Now, obviously we don’t yet have a full season’s worth of data for 2001 but there’s no denying that the trend has changed even more in a positive direction than it has in the previous two years, when it also continued to move in a positive direction.  Let me also remind the reader that in this example, we’re excluding age and park effects, which would in reality be a tremendous mistake if our projections were to be worthwhile.  It happens that in this case, it does not preclude us from demonstrating the point.

Let me take the reader, therefore, on a brief journey into the field of statistics.  When I say “statistics” here I do not mean in the traditional sense as we often hear such as in the phrase “baseball statistics.”  In this case, I am talking about the mathematical field that enables us to describe and analyze a sample of data to reveal something about the reality as a whole.

Let us think of Luis Gonzalez’s real current ability as H.  H does not refer to what he has done to date in 2001 nor does it refer to what he will do for the remainder or whole of the season.  H is the hidden HR% ability that he actually has right now which we can’t define, just yet.  We suspect it’s probably higher than his career-to-date ability of 3.96% but beyond that, we can’t say just how much higher it is.

So before we can go further, we need to actually determine whether our first hypothesis is correct.  The first hypothesis is that his real hidden ability, H, is higher than his career-to-date ability.  To solve this problem, we will employ techniques that you can find in most university statistics texts and that I will attempt to simplify here.

Let us begin with the career HR% of 3.96%.  If Gonzalez’s hidden ability, H, really was 3.96%, then we could look at the 2001 data as a sample of his ability.  A sample in this sense is nothing more than a group of outcomes that we can use to make a determination of the real outcome as a whole, or in this case Gonzalez’s hidden home run ability.

We start by determining a level of confidence, which in this case will be 95%.  I could use a lower level of confidence, such as 51%.  In fact, I have had some of the more mathematically inclined readers write me in the past to argue that I should be using 51% because I will be right more often than I am wrong.  My response is that I would rather include insignificant data in my conclusions than exclude meaningful data.

So with a 95% confidence level, and a theory that Gonzalez’s ability, H, is 3.96%, I can tell you that in a sample size of 82 (the number of times Gonzalez has put the ball in play in 2001 so far), we should get between 0 home runs and 6.7 home runs, 95% of the time.  Gonzalez already has 14 home runs!  So, we could say with 95% confidence that Gonzalez’s home run ability is higher than 3.96%.  In fact, we could make that statement with more than 99.999% confidence if you actually worked it out using statistics!

So, we’ve established that Gonzalez’s real home run ability is probably above his career mark and because we have no reason to believe otherwise, we’ll set the upper limit at his 2001 performance.  Now is where the real fun begins and that is where we begin to zone in on this hidden value of H, which for now we can only say is greater than 3.96% and lower or equal to 17.07%.

There are actually several approaches we can take to solving the problem.  We would have no statistical problem accepting his 2001 numbers on their own except that our knowledge of baseball tells us that if his real ability is 17.07%, he would be by far the greatest home run in history and would have a good chance to go on and hit 80-90 home runs this year… which we don’t expect him to do.

First, we can ask ourselves the question: What would Gonzalez’s real hidden ability, H, have to be for his 14 home runs in 2001 to be within possibility 5% or more of the time (i.e. we want to be 95% confident that we don’t exclude meaningful data).  One other possible solution, which I actually use when updating projections throughout the season, is to begin shrinking the amount of relevant data we consider.  This can be a time-consuming process, however, and that’s where we are able to take advantage of the high speed calculating abilities of a computer.

What we discover is that Gonzalez’s hidden ability, H, would have to be at least 10.5% for us to conclude that he could hit 14 home runs in an 82 size sample (of balls put in play) at least 5% of the time.  Now, to put that in perspective, Mark McGwire’s home run percent in his record-setting year of 1998 was 19.77% and so this new theory is obviously within the realm of accomplishment.

But if Gonzalez’s real ability, H, is 10.5%, as we now theorize for a moment, then that would mean that presuming he continues his relatively normal pace of 3.04 balls in play per game for the remainder of the season, and continues to play in about 129 more games, he would hit 41 home runs the rest of the way and would thus finish with 55 home runs for the season (41 is added to the 14 he already has).

Reminding the reader that we calculate for other factors such as park in age in our analysis, it may ultimately be true that using this 10.5% mark is a good measure of Gonzalez’s newfound power stroke.  But what, if in our analysis, Gonzalez had been almost as good in 2000?  Let’s say he had recorded a HR% of 15% in 2000 in a large sample size?  Would it be fair for us to conclude that Gonzalez’s home run ability must only be at the lower limit of possibility?

Of course we shouldn’t penalize him.  And this is where we reveal quite a bit about how our projections actually come together.

We begin by taking the player’s career numbers, adjusted for a number of factors including his age and park, and make that our “pull direction.”  We want to pull our conclusion in the direction of a player’s career because it is more reasonable to do that than, say, presume his newfound ability is a pure representation of skill.  In other words, as is the case with Gonzalez, I feel more comfortable saying that his real home run ability is less than his great start because his career implies it.  So, in his case, I’m looking for a potential value of H, his ability, that can still yield his 2001 results but that is somewhere between his career ability and his 2001 results.

But to discover this number, I first move back a season.  In other words, if the career hypothesis doesn’t work, and it often won’t, I want to step back one season in the sample.  Indeed, I am actually much more specific when making forecasts.  I usually go back to September of the previous season and continue to work my way backwards.  I add the results from both the old data and the latest data to create a new hypothesis and then test it against all the actual outcomes, in this case using a much higher degree of confidence to allow for natural error as a result of a greater number of samples.  In other words, if I have 20 samples to work with, then naturally one of those samples should fail 5% of the time (i.e. 1 in 20) and so I must use a greater degree of confidence.

Once I find a point at which the hypothesis doesn’t work with all available data, and in Luis Gonzalez’s case, it fails as soon as we go prior to 2001 data, then I search for a number between the player’s career average and the still-considered data to find a value of H that actually works for all the sets of samples and again, I use a higher degree of confidence to make my conclusion, depending on how many sample sets I have.  In Gonzalez’s case, I continue using 95% confidence because I only had one sample set, the 2001 data.

So, having revealed much of how our projections come together, and scaring off many readers who aren’t interested in the details, I change a projection when a player’s recent performance reveals a trend outside what he should be expected to do based on my old theory.  When the old theory no longer works for the latest performance, we see a change to at least an element of his forecast.  For example, I might modify Gonzalez’s home run forecast yet conclude that his ability to put the ball in play remains relatively consistent.

It’s a lot of work and most readers don’t need to understand the reasoning behind what we do.  We consider it a primary part of our job to do the analysis.  It is only coincidence that the first edition of forecasts in May offers the most changes to forecasts and it is because for so many players, the sample size has become large enough for us to make meaningful conclusions.  Indeed, some players take late into the season before they have clearly revealed a different ability.

When we published this response, Luis Gonzalez had just 14 of his home runs in the book.  At the time, we modified our projection on May 1st to tell readers to grab him, projecting him then to finish the season with 55 home runs.  He went on to finish with 57.  Astute readers who listened to us were able to recognize that Gonzalez's home run ability was for real.

Q. What players have convinced you that they’re better or worse than they’ve ever shown prior to 2001?

A. Obviously, the reasoning behind our conclusions is revealed in our lengthy answer to the previous question and so here we can get into the specific players that have shown us something in April outside their previous career accomplishments.  This past week, many players crossed the threshold of sample size that allows us to make new conclusions and so I’ll focus on them in some detail here:

For Barry Bonds to have hit so many home runs thus far, he has proved that his power is better than ever.  Our pre-season projection for Bonds was for a career-high 53 home runs and this one is looking pretty good to start off the season.  Little modification is needed.

Within a month of this, we changed our mind about Bonds as you will see.

J.D. Drew is showing improved power, having hit just 18 home runs in all of last season and having 9 home runs already by May 2^nd this year.  We had originally projected Drew to hit a career-high 24 home runs but have upped his ability dramatically in the newest projections.

Drew finished with 27 home runs in 109 games.

Luis Gonzalez is covered in the earlier answer in this column but needless to say, we’ve modified his power projection dramatically for the remainder of the season.

See earlier.

Mark Grudzielanek already had 5 home runs by May 2^nd after hitting just 7 all of last season.  We have modified his forecast in the newest issue and he is an excellent bet to finish with another 5-7 home runs the rest of the way, which gives him a shot at setting a modest career high.

Grudzielanek hit 8 more home runs and achieved his "modest career high."

Gabe Kapler, who had stolen 21 bases in 885 career at bats prior to this season, has already tried to steal 4 times and has just 28 at bats thus far in 2001.  Look for Kapler to steal 10-12 bases the rest of the way and finish with a career high 15 or so.

Kapler did even better than we thought he would, achieving a career 23 steals.

Aramis Ramirez’s power is for real and I expect him to get 15-20 more home runs the rest of the way, in addition to the 7 he had as of May 2^nd.  That would give him 20-30 home runs for the season which would be a tremendous improvement over his 12 home runs in 561 career at bats before the 2001 season.

Ramirez finished with 34 home runs.

Reggie Sanders’ home run stroke is looking better than ever.  Though I still consider him an injury risk, I could see him doing something crazy, like hitting 25 home runs in 300 at bats the rest of the way, which would give him 30-35 for the season based on where he was on May 2^nd.

Sanders finished with 33 home runs in 441 at bats.

Chris Singleton’s running game has graduated to a new level, which is too bad because I really don’t think he’ll play as much as he should.  He stole over 20 bases each of the previous two seasons and if he could get in the lineup everyday the rest of the way, he would steal at least 20-30 the final four months.  I just don’t see him playing enough to do that but his ability is looking different and is worth noting.

Singleton finished with 12 stolen bases.

Albert Pujols is for real and is off the scale compared to our original expectations.  Look for him to play regularly the rest of the way, hitting .260-.265 with about 20 home runs yet to be hit.

Though we started to pick up on Pujols ability at this point, he was still a dramatic miss in the projections and we admitted that.

Toronto’s Kelvim Escobar has become a dominating reliever and our newest projection reflects that.  I expect him to pitch another 55-60 innings, striking out maybe 75 and walking about 20.  Buck Martinez says that he’s not going to be the closer anytime soon but he’s an interesting one to watch for 2002 or further down the road.  He definitely has closer stuff.

Escobar remained effective but Buck Martinez later moved him into the starting rotation, making him even more valuable to fantasy leaguers.

Kerry Wood is more dominating than ever, striking out 46 batters in his first 29 innings of 2001.  Look for Wood to average about 7 strikeouts per start the rest of the way, which might sound low but would add up for a guy that should go 5-6 innings per remaining start.

Wood finished the year with an average of 7.75 strikeouts per game and just over 6 innings per start.

Speaking of strikeouts, the Orioles’ Willis Roberts struck out 27 in his first 27 2/3rds innings this season.  Look for Roberts to keep striking out hitters but to get hammered the rest of the way.  I could see him getting maybe five starts and a handful of relief appearances and getting knocked around.

Roberts did get knocked around, ending up with a 4.91 ERA.  To his credit, he did finish with 9 wins.

John Lieber is for real.  Look for him to keep his ERA under 4.00 this season.  Ditto for Wade Miller, who has been stunning so far.

Lieber had his best season ever, going 20-6 with a 3.80 ERA.  Miller finished with a 16-8 record and a 3.40 ERA.

From AskDL June 10, 2001:

Q. Do you think Barry Bonds will challenge Mark McGwire's home run record?

A. There is enough data in the books now that I am convinced that if he stays healthy, Bonds will not just challenge McGwire but he will shatter the record.  Before the season began, we had projected Bonds to hit a career high 53 home runs in 508 at bats and it was the subject of several emails of disagreement.  As it turns out, we were wrong.  Bonds is even better than we imagined and we're not dealing with the same player here.  Cross your fingers that he stays healthy if you want to see one of the greatest offensive seasons of all time.

Though it seems easy to have said this, remember that we published this in early June and we were fairly emphatic that he was going to break the record.  Bonds, as everyone knew, finished with 73 home runs.  It's not to our credit as it took us until mid-May to up his projection from our opening day forecast of 53 home runs (which was still a career high forecast) to numbers that consistently fluctuated from 71-80 after this comment was posted.

Q. Luis Gonzalez continues to hammer the ball.  How many home runs do you see him finishing with?

A. Like Bonds, we have an entirely different player this year.  As we went to press, Gonzalez already has 25 home runs.  Look for him to hit another 30 or so the rest of the way and finish with about 55 homers.

This was essentially a repeat of the forecast we made at the beginning of May and, of course, Gonzalez finished with 57 home runs.

Q. I have a bunch of pitchers in my league that are off to bad starts, including Barry Zito, Gil Heredia and Rick Helling.  Should I jump ship on these guys?

A. Zito is a far better pitcher than he has shown and has relatively strong job security given the current mode the A's are in.  As for Heredia, he too is a far better pitcher than he has shown but unlike Zito, doesn't fit into the longer-term plans that Oakland obviously must be considering.  The problem with Heredia is that he could be a good major league pitcher still that gets cut a la Mark Portugal and then never finds a team.  I would simply hang on to Heredia until he gets cut and then hope he finds another team because, as I say, he's better than he has looked.  As for Helling, I see him posting a high 4's ERA the rest of the way but also picking up 5-10 wins so you'll have to decide if that fits into your plans.  I would try to move Helling if you can.

As everyone knows, Zito went on to have a dominating second half.  Heredia lost his spot as a starter.  Helling finished with a 5.17 ERA and 12 wins.

Q. Johnny Damon has literally killed me this year.  I know he can't be this bad for real but just what is his real ability?

A. Despite his terrible start to the season, I see Damon as a .290s type who will hit maybe 5-10 home runs the rest of the way with 20-25 steals.  Adding that to what he's already put in the books and I'm not a believer to the theory that he will finish with a .300 average just because he usually has.  A .300 hitter can't simply hit .400 in order to finish a season with their "normal" numbers and so I see Damon hitting .290-something the rest of the way, thus finishing with a .260-.265 average or so.  The ironic part is that I would then say he goes into 2002 with an excellent chance to hit .290 that year.

Damon finished with a .256 average, 9 home runs and 27 steals.

Q. I have Chris Michalak in my league and he has been carrying my pitching staff.  Is he for real?

A. Michalak is so likeable by virtue of his story.  Here is a 30-year-old rookie who bounced from organization to organization looking for an opportunity.  Heck, he didn't even become a starter in the minor leagues until the 2000 season.  Six teams gave up on him and now he's the quiet co-ace of the Blue Jays' staff, along with Chris Carpenter.

Now that I've hyped him, let me tell you: Trade him quickly.  Trade him while he still looks this good.  Though I continue to root for his success, he is a soft-tossing finesse type without much movement on the ball and I expect him to get hammered hard soon.  I could see him posting a 7-8 ERA over 5 or 10 more starts before the Blue Jays figure out why he wasn't good enough to stick with Tampa Bay, among others.  Because I like underdog stories, I hope I'm wrong.  From the forecasting perspective, I have to say: move quickly.

Michalak did indeed collapse and was ultimately released by the Blue Jays, was claimed by the Rangers and then continued to get hammered by hitters.  The question/answer is shown to illustrate the importance of not playing too much weight on players whose statistics look good but who really don't have the stuff to get hitters out. 

From AskDL January 1, 2002:

Q. Which Johnny Damon is the real one: 2000 or 2001?

If a hitter was "really" a .300 hitter and you gave him 644 tries (or at bats) to get a hit, the science of statistics tells us that there's about a 2.5% probability that he will get fewer than 170 hits over that sample, making him look worse than a .264 hitter, even though he's not.  Damon hit .256 last year and I will say that we notice him because he was one of those rare cases who experienced results that do not reflect his real ability.

He's going to like playing in Boston as the park is conducive to his hitting style, more even than other sluggers who take advantage of the Green Monster.  Look for Damon to bounce back by topping .300 with about 10 home runs and 30 steals.  The power-speed numbers won't be much different but his average should revert to where it was two years ago.

From AskDL February 13, 2002:

Q. Will you be making forecasts for the new Japanese players and can you tell me any other Japanese players who are good enough to play in the majors right now?

A. The next update to our projections this upcoming Sunday will display the first forecast we have created for So Taguchi of the Cardinals.  He wasn't an extraordinary player in Japan and I see him as a .270s type in the majors with little power and speed.  Even if he played a full season with St. Louis, which I don't think he will, he would be a 5-10 home run type with 15-20 steals.  Don't count on the speed though because he's not going to play.  I suspect that Taguchi will go relatively high in some fantasy drafts because some fantasy owners will make the false conclusion that any hitter from Japan is the next Ichiro.  Ichiro was a great player in Japan, Taguchi was not and that's the way both will be for the long-term in their major league careers.

We have not yet created a forecast for pitcher Kazuhisa Ishii, who officially signed with the Dodgers a few days ago.  It's worth considering that in Japan last year, Ishii was the eighth-best Japanese pitcher in terms of ERA (3.39) but did have some control problems (82 walks in 175 innings).  To put his performance in perspective, Hideo Nomo had a career 3.15 ERA in five years in Japan, Kazuhiro Sasaki had a 2.31 career ERA and Hideki Irabu had a career 3.41 ERA.

Among other Japanese players, if I had to name a single one who would be a great major leaguer, it would be a name that almost no one knows in North America but will be a household name a few years from now if he eventually plays in the majors.  Twenty-eight year old Hideki Matsui, who won the Japanese batting title last year, is an outfielder for the Yomiuri Giants.  Matsui is just peaking now and could hit .340 in the majors with 25-30 home runs, 100+ RBI and almost as important to major league teams, a tremendous number of walks, 120 or more.  If your league allows you to draft unsigned Japanese players, regardless of whether they're signed with a major league team, Matsui is a great player who will eventually command huge dollars in the majors and will exceed expectations when/if he does come to North America.

For what it's worth, this past December Matsui signed a one year contract extension with the Yomiuri Giants worth $4.7 million US, giving him the highest single-season Japanese baseball salary ever.  Look for him to make a move to North America at the end of the 2002 season.

Q.  Do you put any weight into a manager's comments about players?  For example, Joe Kerrigan remarked that Nomar Garciaparra and Trot Nixon should be ready to accumulate 30 steals in 2002.

A.  I answered an almost identical question regarding Sammy Sosa two years ago, when then new manager Don Baylor remarked that he wanted to see Sosa steal 30 bases and many readers wrote us to complain about our 7 stolen base projection, saying we were disregarding the manager's comments.  Sosa went on to steal exactly 7 and no one remembered Baylor's pre-spring comments after the season.

A player's talent does not change by what a manager says he wants the player to do.  There are occasions where a manager's overall approach will be a factor and Tim Johnson, in his one season as manager of the Blue Jays, did coax 29 steals out of Jose Canseco, the second-best total of his career.  However, if we examine the Blue Jays of 1998 more closely, we discover that Johnson raised the running game for the entire team.

So, to answer your question, we do take into account a managerial style in terms of projecting steals as a whole.  Managers comments can also dramatically influence some categories, such as saves forecasted for relievers.  Jimy Williams, Kerrigan's predecessor, did prefer to run his players a bit less frequently, partially because he didn't have a lot of speed to work with but more because everyone was getting caught stealing (the Red Sox as a team finished with the worst SB% in the American League at a terrible 57%).  Joe Kerrigan, after taking over for Jimmy One-M, did seem to have a good sense of when to run the players and under his management, the Red Sox showed a much better ability to run.  In that respect, we've already considered this in our forecasts but I don't see either Garciaparra or Nixon suddenly becoming great base-stealers.  After Garciaparra gets caught by a wide margin a couple of times, I imagine Kerrigan would rethink such comments.

From AskDL February 24, 2002:

Q. To what extent does batting order influence your projections?

A. It affects it greatly because even after we've created our model of a player's skills, we need to simulate the season at high speeds many thousands of times to begin building a realistic base.  The model takes this into account and I should remind you that just because a player bats leadoff does not mean he will not get many RBI and the converse is true for cleanup hitters.

I cannot emphasize enough that though we take the projected batting order into account, we also recognize and consider that a batting order changes.  A few years ago, many readers complained about our Cliff Floyd projection because we projected 90 RBI when he would be hitting out of the leadoff spot.  Floyd started the season in that role but by season's end had been dropped lower in the order and went on to drive in around 90 runs.

Q. Are you considering the sudden revisions to players ages, especially among players from the Dominican Republic who are having problems renewing their Visas?

A. Yes, every time a change such as that comes to light (such as with Bartolo Colon this week), it is automatically adjusted in our model and will then be considered when we revise projections in the next update.  Fortunately, thus far, none of the corrections in age has been significant enough to dramatically affect forecasts but it is entirely possible that others will come to light that do change a projection by a wide margin.

Q. How can you value players for fantasy leagues at $50 or even $60?  There's no way that players will go for that kind of salary in my fantasy league. 

A.  You're right.  They'll never go for that kind of salary.  If you are a member using our interactive forms, the value listed is projected value and not salary.  We're trying to tell you what each player will be worth to your team, not project what price they will sell for in the auction.  There are many players who will sell for $1 who will be worthless and there are other players who will be worth $20 or more who will sell for much cheaper than that.  Remember, the values you see are projected value, not projected selling price!

From March 3, 2002:

Q.  I noticed a number of players have a significant increase in projected slugging percentage.  Is this based on statistical reasoning or scouting?

A.  It's a combination of both.  For every player, we have a scouting rating for their ability to make contact (which has a huge influence on singles), hit for power (which obviously impacts home runs and to a lesser extent, doubles), hit to to the gap (which affects both doubles and triples) and run the bases with speed (which affects triples and is not necessarily related to the ability to steal bases).

We have considered publishing these ratings but at this point, they are so dynamic and "un-user-friendly" in format that it's just not possible at this time and beyond that, there is a point at which our information is private in that they reveal industry secrets in our forecasting methods.  Also, in a large number of players, we don't have this sort of scouting data available and therefore, attempting to publish it would not be possible unless we leave such gaps, which would certainly lead to confusion that we could not respond to at this time.  The projected statistics for each and every player speaks for our conclusions and one could infer, with some work, exactly what we think of each player.  If we eventually do publish these ratings, they will obviously be added to the member's section that already exists.

Once we have the scouting data compiled for each player, we compare it to our results model.  If, for example, that all A+ power hitters with D contact ability went on to hit 20 home runs per 550 at bats in previous years, we use that and are constantly tweaking the results to reflect changes in the major leagues, such as the new strike zone last year.  Thus, a new player with A+ power and D contact ability might, as a base, get a 20 HR projection if our historical results typically yielded these numbers.  Slugging percentage is obviously a simple statistical calculation that combines singles, doubles, triples and home runs and thus, is never a category we actually projected directly.  A high increase in slugging percentage may not have been our intention but if the scouting data leads us in that direction, we can't ignore it. 

In the case of players for which we do not have first-hand scouting reports (and we never use any other scouting sources other than our own), we use more of a statistical approach.  Sometimes, we have limited scouting data and I prefer to see a player at least 5-10 times, at minimum on video if not in person, before assigning a scouting rating.  Therefore, players without scouting data are typically low-ranked players who have never appeared above the low minors or returning major leaguers who I haven't seen in the previous while (such as players returning from Japan).  In that case, our numbers use more of a statistical approach where we use our private tables for translating minor league performance and other league performance.  Sometimes, the statistical translations have proven to be the most accurate methods and so, we will often ignore scouting data if statistical evidence is so overwhelming that it outranks personal observation.

Q. Do you recommend trying to draft a balanced team or drafting the best available player and trying to deal for areas of weakness later in the season?

A.  I have always advocated drafting the best available player for so many reasons.  First, this isn't real baseball here and the players don't have to actually field the position.  In real baseball, you have to field a second baseman and a shortstop and they really must be able to play the position with some proficiency.  Indeed, I can honestly say that no simulation game in existence truly reflects the real importance defense has on the game of baseball.  If I were a real GM, I would be considering position scarcity, etc... In fantasy baseball, it's a really good way to end up with some great hitting infielders and a mediocre team.

As for trading later, obviously every team should be trying to do this no matter how good your players are.  Believe me when I say, go for the best available remaining player, regardless of position, and you'll end up with a better fantasy team.  As it happens, A-Rod is going to be a top player in virtually every league anyway and he would be that if he played first base or the outfield.  That he plays shortstop just makes him that much more of a bonus when he ends up on your roster.

Let me add that I have won many leagues with relatively lousy players at catcher, second base, third base and shortstop.  It's worth considering that most teams will have bad players here and if you do try to factor in position scarcity, what happens is that you get a great second baseman and then can't afford to buy a Barry Bonds or Todd Helton.  That other team who shelled out the bucks for Bonds or Helton but settled on an Alex Gonzalez at shortstop or Carlos Febles at second base, has better numbers than you do and you can't beat him.

Q. I am worried about dumping some great players at over their projected value (e.g. Manny Ramirez and Ray Durham) because of draft inflation, which is definitely evident in my league every year.  Can you propose a strategy to deal with this?

A. Draft inflation is a reality and if your league has plenty of great keeper lists, clearly there will be more money available to spend on players than talent available.  Having said that, the 75-80% target rate, discussed in recent articles on our site, makes it unlikely that draft inflation will actually occur before the auction begins because though the keeper lists will be undervalued, they won't be as low as 80% in a combined total.  This is because there are always plenty of overpriced rosters, relative to this 75-80% goal and thus, you can go into the auction with your typical target, regardless of draft inflation.

I know it's hard to let players go and the best solution to this is to find a couple of dramatically underpriced players through a trade with another team.  This can allow you to fit an overpriced player onto your roster while still working toward your goal of underpaying the team.  It's nice to keep overpriced players but in reality, you're trying to win and if these guys aren't going to help you achieve that goal, then you have to let them go.

Another option, of course, is to trade these players as there is a team in every league willing to pay well to get a Ramirez or Durham.  That means you can trade a $45 Ramirez or a $25 Durham for a lesser known player who will give you a greater return on investment, such as a $10 Kelvim Escobar or a $20 Billy Koch.  This approach is usually the preferable alternative but if you can get a few real bargains on your roster, you can often squeak in a few slightly overpaid superstars that you would otherwise be reluctant to let go back into the free agent pool.

Q. My ranking sheets have Randy Johnson projected to be worth $75.  Even if I bid just 75% of projected value, I will be setting a league record for the highest salary ever.  Should I do that?

A.  Of course not.  When the highest possible salary you can imagine is lower than your calculated bid, as discussed in our recent essays, then you have to pick up the player for $1 more than you expect anyone will bid for him.

Also, don't be afraid to miss picking up a player by underbidding.  Do your best to estimate what your opponents will bid for the player, add $1 or the minimum bid, and give it a shot.  If he goes elsewhere, then recognize that you missed the boat and move on.  If it happens often, you'll get burned if a player gets "stolen" for $51, then you just have to imagine that he was protected at that salary and forget about it.

Having said that, there are players, such as Johnson, that would give you such a good return on investment that you can safely bid higher rather than lower and have a pretty good chance of being happy with your purchase.  I will tell you that in many leagues, I would not be afraid to bid $45-$50 to get The Big Unit, though he is an unusual exception.  Players such as Johnson can help you win three or four categories almost outright (depending on how many categories your league uses) and the only real reluctance you understandably should have is that if he gets hurt, your season goes down the drain.  Of course, if your highest paid player cost you just $35 and he gets hurt chances are you're not going to win either.

From AskDL March 19, 2002:

Q. In your projections, Dave Williams takes a major swing up in home runs allowed (about double what he allowed last year).  Can you explain?

A.  Yes, we're projecting Williams to struggle in 2002, largely because of the way we consider his minor league performance in the equation.  Though he allowed just 15 home runs in 114 major league innings last year, he allowed 11 home runs in just 69.1 minor league innings last year and we can't ignore that.  I just don't find it believable that he instantly improved his ability to prevent home runs upon getting called up to the majors.

As for the rest of our projection, he walked too many hitters, struck out too few and he's a good example of how a statistical performance doesn't always represent ability.  He has ordinary stuff, at least at this stage of his career, and to expect him to have an ERA under 4.00 at this level, with the way he throws and the relative lack of control, just doesn't seem reasonable.

Q.  Jason Marquis's forecasts continue to take a beating.  Any reason?

A.  Marquis is a great example of how spring performance can affect our forecasts.  Though it's not a purely statistical observation, Marquis doesn't look good in camp and up until a few days ago, his spring ERA was almost 15.00.  Statistically speaking, it takes a dreadful performance in spring training to convince us that the player we saw last year isn't the player we'll see this year and in Marquis's case, he crossed that line.  He is, without question, one of the earliest true modifications to our 2002 forecasts, at least in terms of performance.  Usually, it takes until around May before we can modify a player so drastically but even with the modifications, we're still projecting a half-decent year, at least in this era (his current forecast has his ERA just below 4.50).  Don't be misled to believe that all 15.00 ERA spring pitchers get a modification.  There are other factors that we consider when making such adjustments and in Marquis's case, he has the signs of a pitcher for whom an immediate adjustment was necessary.

Q.  My draft is on a Friday and you make your projections available on Sundays.  Can you change the day for me?

A.  This is such a common question that I thought we should include it in an at least one issue here.  I would estimate that each week, we get at least 10-20 requests asking us to change the publication date to a specific day of the week and in fact, I believe that in the past couple of weeks, we have many votes for each and every day between March 20th and April 15th!

A few other sites claim to update their forecasts daily but what they typically do is "tinker" with the projections in that if a player gets hurt, they modify that individual player but don't look or modify the major leagues as a whole.  If Pedro Martinez gets hurt, this must affect each and every hitter in the American League because now instead of Pedro, the hitters are facing an inferior pitcher in his place.  It may not seem like much but if you remove Pedro Martinez from the American League, you must modify every American League hitting forecast by about a third of a hit or more, accounting for the difference between Pedro and the average major league pitcher that might replace him.  Though a third of a hit doesn't seem exciting, combine this with changes to about 30-40 players daily (whose role or status changes at that pace) and you see that if you really want to have great projections, you need to do these sort of adjustments.  It's one of our not-so-well-kept secrets that contributes to projection accuracy.

This is the sort of forecasting we do and all of our forecasts are interrelated with each other.  When we change one player forecast, every other forecast gets impacted it and that's why with each change, even the players whose role has not changed will show the slightest differences from the previous version of the set.  It takes us almost a full week to do this sort of modification and the attention to detail pays off with the accuracy of our forecasts.

We picked Sunday based on a reader poll taken in 1998 that seemed to tell us that, especially during the season, readers would like to have fresh numbers on Sunday morning because most leagues have a Sunday night weekly cutoff for transactions.  Unfortunately, once our publishing schedule has been published (and ours was published in December), we cannot and will not make changes to it as many readers plan their schedules around it.

As our site continues to expand and grow, and it continues to do so rapidly, we will expand our infrastructure to perhaps produce daily updates.  We did daily updates every day of the season during the 1999 season and I must admit that not only was it tasking but our forecasts, though still great, did not then have the sort of connection to each other that they do now.

We continue our research and development and it is entirely possible that in future years, we will have more frequent updates as part of our ordinary schedule.  If the cost to do so turns out to be prohibitive, we could always make such updates a paid option for full members but at this point, we're not ready to do more frequent updates, no matter how much this would be worth to our readers.

Q.  Why do you have so few at bats projected for Morgan Ensberg?

A.  The Ensberg forecast is because I don't think he's ready to play in the majors and I believe that will show quickly.  He's a good example of how the plan can go out the window if a player struggles (which we also say will hapen to Cesar Izturis of the Dodgers).  I'm projecting Ensberg as a .230s hitter with good but inconsistent power.  Ensberg has just 4 major league games under his belt and though he's been exceptional in the minors the past two years, I don't believe it will immediately translate and he's older than most think at twenty-six.  If I'm wrong about his playing time, it will be because I'm wrong about his ability.  If he goes out and hits .300 in April with the same sort of power I know he has, then obviously he's not going to lose his job.  As it stands now, it would be contradictory for me on one hand to project him as a .230s hitter but, given that the Astros have several other candidates even without Chris Truby, also say Ensberg is going to play 150 games.  I suspect that if you disagree with me on him, it's because we see a different type of player.  I say he's going to lose whatever job he wins if he hits as I expect.  If I'm wrong about him, it will be on the ability side.  The playing time simply goes in hand with that.

From AskDL March 26, 2002:

Q.  How does making the Opening Day roster affect your forecasts?

A.  It affects it dramatically because by virtue of the percentages, Opening Day players have a much greater chance of getting playing time than of those who initially start the season in the minors.  Of any projection set other than those we made way back in January, our projections will change more this weekend than at any other time this pre-season.  Literally, as the Opening Day rosters are finalized, we have to carefully revisit each and every one of the 1,200 players and evaluate their real prospects for playing as much, more or less, than we've previously published.  Indirectly, this can affect the actual abilities too because if we find that the quality of pitching, for example, will be higher as a whole in the league than previously thought, it can have a marginal effect on the hitting forecasts and vice versa.  We already do a thorough review each week but the final week of spring training gives us more data than we've previously had and so, look for dramatic changes all around.

Q.  Why do your interactive ranking forms not rank pitchers between starter, closer and setup man, like you break out hitters at each position?

A.  I actually answered this question a while back but it continues to come up and for newer members in particular, it's worth covering again.  For hitters, we let you specify how many games a player must have played at a position to qualify for that position in your fantasy league.  For pitchers, there are a hundred different ways these pitchers get qualified.  Take the example of Kelvim Escobar.  Escobar pitched in 59 games last year, started 11 of them and had 6 saves.  By the saves standard, he can't qualify as a "closer" from last year but that's his expected role for 2002.  In some leagues, he would qualify as a starter and in others he would qualify as a middle reliever or closer. Especially in the case of pitchers, leagues that attempt to say that a pitcher qualifies as a relief pitcher or starting pitcher tend to use non-statistical criteria.  They simply draw up a list and say, these guys qualify as starters and these guys qualify as relievers.  There is often no rhyme or reason to it and for that reason, we cannot break pitchers out into different "types" of pitchers. 

Q.  I used your interactive forms and now I'm stuck with Ishii at $45.  What should I do?

A.  Well, that wasn't a smart play.  The interactive sheets give you an estimate of the projected value a player will have to your team.  They are by no means intended to be an estimate of what you should bid and in fact, the form advises you at the top that you want to assemble a team that costs only 75-80% of its value. Neither does that mean you should even bid 75-80% on players because in some leagues, Randy Johnson might be "worth" $100 but you shouldn't bid 75% of $100! 

In some cases, you might bid a bit more to get the player and in other cases, much less.  You can't go blindly into an auction deciding that you are going to bid what you think a player will be worth or 75% or whatever.  You need to maximize your returns and the way to do that is to bid exactly $1 more than anyone else would be willing to pay, if that value produces a desirable return on investment.  That's where your knowledge of your opponents must come into consideration.

Since the forms are in descending order of projected value, you should frequently find yourself skipping a top ranked player, for whom there is little or no interest among the competition, and taking a lower ranked player, leaving a space open for that higher player.  Return later in the draft to pick up an Ishii and you'll get him cheap.

Let's pretend for a moment that we could fast forward to the end of the 2002 season and suppose we could produce the final ranking forms for your league, based on what players ultimately turned out to be actually worth.  Do you really believe that you should draft in descending order of what they were worth?  Let's pretend that at the top of the list of 2002 value, someone like light-hitting utility infielder Chris Woodward hit 60 home runs and stole 60 bases.  No one would have predicted that.  Let's also say that Randy Johnson went on to win 25 games and in your league was the second most valuable player.  Now, rewind to draft day.  Would you take Woodward first if you knew this was going to happen?  Not in a million years!  You wouldn't need to take him first.  You would take Johnson first and you would wait as late as possible to pick up Woodward, perhaps leaving a slot free on your roster for him but not drafting him in the early stages of the draft.  This way, the second tier of players you get is superior to what it otherwise would have been.  The same applies to using our ranking sheets.  Even if we're right, you shouldn't always take the highest ranked available player if you could get him later in the draft.

Q.  Would you say that your approach to forecasting is a stats-based approach?

A.  Though that would be fair, I've talked in the past about how much of our forecasting is related to personal scouting of the players.  I think it would be accurate to say that there is a greater statistics emphasis in our forecasting than raw scouting and if anything, that's because we are trying to forecast statistical results.  Occasionally, I get a note from a reader who will say that they disagree with a forecast, asking why I said Player X would not be good.  Good and bad mean different things to different people and if you think a forecast of an ERA of 7.00 is bad (and it would be safe to say that is "bad" for most teams), then we are making an implied non-statistical observation about a player.

Particularly when it comes to emerging minor leaguers, I am a strong advocate of translating their minor league success or lack thereof, to a major league context.  Here, we have information that tells us precisely what a player has shown he is already able to do.  All the scouting in the world can't overcome that if a player, for whatever reason, is unable to put it all together, his talent means little.

I can look back on a decade of forecasting now and see that we've been particularly strong at recognizing that minor leaguers weren't ready, time after time.  Those are usually the players that get the most disappointed response from readers before a season is played.  Ironically, most of those have been the result of statistical analysis superseding my own scouting observations and so, we will continue in that tradition for the foreseeable future.

As one final example, back in 1998 and 1999 I had scouted Roy Halladay up close perhaps more than any other top prospect in baseball at the time.  Looking back in my scouting notes from that time I wrote "extraordinary movement" and "incredible stuff" in my notebook.  I had not seen a young pitcher with such great potential.  During the winter prior to the 2000 season, I performed our statistical analysis on his minor league career, using our equations and methods for translating what he had done in the minors to what the trend implied for the majors.  The analysis said that Halladay would collapse in 2000.  I spent many long hours double-checking the equations and the results in our system and I could simply not understand how a pitcher with such great potentia and stuff was showing such a terrible initial forecast for 2000.  After much temptation to override the statistical analysis, I ultimately and reluctantly published a disastrous forecast for Halladay prior to the 2000 season, the result of pure statistical analysis that ignored my opinion about his ability.  His stuff said he should be a great pitcher but the translated minor league results said that he would be one of the worst pitchers in baseball 2000.  Halladay went out and had a terrible season, going 4-7 with a 10.64 ERA in almost 70 innings.  The stats had superseded my own conclusion and, at least in this case, ultimately proved to be right.

I tell the story not to show that statistics are a more reliable method of forecasting as that would be a hasty generalization based on a single example.  I use this example because, as we head into 2002, I'm doing exactly the same thing for Roy Oswalt.  The stats and our methods of analysis tell me that something different than I have observed, in terms of his ability, is about to occur.  Perhaps this will prove to be an example that shows the value of scouting over statistics.  Time will tell.

From AskDL April 5, 2002:

Q. If a player hit two home runs on Opening Day, such as Jacque Jones, that puts a considerable dent in the 15 home runs you've projected for some players to have this year.  Does that mean you will now list 17 home runs as his projection?

A.  First, if you're asking whether I've changed my mind about Jones, no I haven't, but he's still a decent player who should hit in the high .270s or low .280s with some power and limited speed.

This question creates an opportunity for me to address just what the forecast means.  If we project a player to hit 15 home runs and he hits two on Opening Day, it does not necessarily mean that we will upgrade his forecast nor does it mean that there's a "dent" in the forecast.  Rather, a 15 home run player can hit two home runs and still be a 15 home run player.  Now, if he really is a 15 home run type and already has 2 home runs in the books after one game, chances are greater that he will finish with 17 home runs rather than 15 home runs but still actually be a 15 home run player.

Let's use the example of flipping a coin.  If I said that 100 tails would come up in 200 flips and we flip the first five coins and get 5 straight tails, it does not necessarily mean that I will upgrade the forecast for the remaining 195 flips, though I might eventually conclude, if the trend continued, that we have a different type of coin than the fair one in which I originally believed.  I might still be convinced after five straight tails though that we're dealing with a fair coin and then forecast that 50% of the remaining flips will be tails.  Now, the implication of such a forecast would be that, because 5 tails are already in the books, that the end result is now forecasted to be greater than 50% tails because my 50% projection for the remaining flips can't prevent that we already have a 5-0 head start for tails when the projection is revised.  The fair coin doesn't remember that it already came up with five tails but it does clearly affect the final totals of our 200 flips.

The distinction I wish to make to readers here is that our forecasts, at any point in time, is the forecast of what the player will do for the remainder of the year.  This in fact should be the only concern of a reader.  If a player already hit 2 home runs, you can't get those 2 home runs on your fantasy team, for example, because they're already hit.  Moreover, if a player has already hit 2 home runs and we projected 15 originally, it does not mean that 2 of those 15 have been hit.  Rather, it means that the player has 2 home runs and that may or may not convince us to re-assess the player's skills.  So, you could easily see a projection for Jacque Jones in this weekend's update, that he will hit 14 or 15 home runs the rest of the way.  Given that he has already shown good power, those totals are already in the books and obviously improve his likely seasonal totals but do not necessarily convince us that we're wrong about his remaining seasonal ability.

From AskDL April 13, 2002:

Q.  What do the values on the interactive sheets mean at this point?  Are they the dollar values over the remainder of the season or are they the dollar values including the actual numbers so far?

A.  At any point of the season, the interactive ranking sheets in our member's section are only the projected relative values for the remainder of the season.  Our goal is to create winners and thus, it does no good to tell you that a player is off to a hot start, unless it impacts our forecast.  Therefore, the interactive ranking sheets tell you, based on all factors we've considered, what the player's expected relative value is for the remainder of the season.  Many members use these forms to evaluate trade ideas and such once the season gets underway and though the forms remain active, I personally believe the forms are better suited for pre-season use only.  Once the season is underway, you are much better to use the individual player projections, numerically that is, and evaluate two or more players beside each other to see how they impact your precise team.  Of course, we leave the forms online regardless and they will project remaining value throughout the season but I personally think it's better to think in terms of the actual projected performance, once the season begins, evaluating how each player impacts your specific fantasy team.

From AskDL April 21, 2002:

Q.  Do you believe in Linear Weights and if not, why not?

A.  Bill James recently published a book called "Win Shares" which articulated his feelings (which also happen to be in common with mine) about the flaw in the Linear Weights method for valuing players.  As James correctly observes, each real major league team does not start with 81 wins and then increase or decrease based on whether a player is above or below average.  Rather, as James argues, the bottom tier of baseball is filled with the top echelon of professional baseball players, who are good enough to make the majors.  As he demonstrates, it is unfair to say that a .510 quality player is marginally different from a .490 quality player.  Though I have always admired James' writing, I do recommend this book only for the most serious of Sabermetric readers as it contains more statistical analysis than usual and James spends a great deal of time explaining the reasoning behind his methods, only some of which I happen to agree with in this case.

Beyond the problems with Linear Weights and my own general disagreement with the concept, statistically-inclined readers who have not tracked down the book in which the concept first appeared would be well-served to search used bookstores or online auctions (to the best of my knowledge, it is long out of print).  "The Hidden Game of Baseball" by John Thorn and Pete Palmer is a gem of a book.  It explains the early stages of the Linear Weights system but the parts that, in my opinion, hold the most merit are those which critically evaluate the so-called "book" in baseball.  Thorn and Palmer argue for example that the sacrifice bunt is almost never a good play despite how often it is used and they also look at other in-game strategies such as the intentional walk and the batting order.

So, the short answer is that, no I do not agree with using Linear Weights to measure player value because I do not believe the quality of players should be measured against the average but I do think that Thorn and Palmer have contributed a great deal to the study of baseball and I strongly endorse many of their other philosophies about the game.

From AskDL April 28, 2002:

Q.  I am wondering how you can project more than 20 home runs for Matt Stairs in less than 300 at bats.

A.  In our latest forecasts, just published today, Stairs is projected to hit 19 home runs in 271 at bats the rest of the way, which is a very good power ratio.  He is/was a difficult projection to make because I believe his power is much better than his totals from the past two years imply.  In fact, with the Cubs last year, 12 of his 17 home runs came on the road and in the previous three seasons before this year (two of which were in Oakland), Stairs hit 47 home runs on the road and just 29 at home.  Last year, Miller Park was a great home run park for left-handed hitters, increasing lefty home runs by more than 30%.  So, that's a piece of my reasoning behind the forecast but his latest one is downgraded slightly to reflect that another week is off the schedule and that Stairs has struggled so far.  I would like to see him get more playing time but that doesn't appear in the cards.

From AskDL May 6, 2002:

Q.  I have focused on acquiring players who play multiple positions so I can cover different positions in case of injuries or off days.  Is this a bad strategy and should I be more focused on player performance rather than how many positions he can play?

A.  I don't recommend focusing on position eligibility for more reasons that I can count.  Firstly, most leagues have some sort of allowance to replace an injured player, whether that means you have to pick the "best of the rest" from the free agent pool or you have to promote someone from your farm system at the end of the week.  If you'll be forced to pick from the free agent pool, for example, you might end up with a player who is just as good as the five position, .200 hitter that you're keeping on your roster for so-called flexibility purposes.

There are players who are fortunate enough to qualify at more than one position and contribute decent fantasy numbers.  For example, in some leagues, Tony Batista qualifies at both third base and shortstop but it wouldn't make me much more more likely to draft a Batista if he only qualified at one position.

So, my advice is this:  Focus first on performance and consider it a bonus when you end up with players who, by virtue of their flexibility and your league rules, qualify at more than one position.  Unless your league absolutely prevents you from ever replacing an injured player, it doesn't do much good and even if that is the case, then you still have a hole to fill when you move your so-called utility guy around.  It doesn't solve the problem of vacancy so much as it simply moves it from position to position, which can be handy but isn't a reason worth giving up the big offensive numbers.

There are players such as Batista who qualify at multiple positions and who have some decent fantasy baseball value.  Mark McLemore, for example, qualifies in most leagues at outfield, third base and shortstop and also steals plenty of bases.  If you absolutely must have flexibility on your roster, and I don't believe you must, then at least go after players like these.

From AskDL May 19, 2002:

Q.  When, if ever, should I begin to change the %'s from 65/35 in hitting to something closer to 50/50?  In other words, is the 65% hitting the goal during the draft or all season?

A.  I strongly recommend that you should always be thinking in a 65/35 hitting/pitching split and I've acknowledged many times that there are as many pitching categories as hitting categories in most leagues.  The reason you must think this way is because everyone else does and thus, you would be overspending.  The way to think about it is this way:  There are actually two fantasy teams which you are managing, one of which is made up of hitters and one of which is pitchers.  They are separated in such a way that other than because of the way you allocate salary, it is not possible that the performance of one group can affect the other, unlike what happens in real baseball (for more reasons worth mentioning, starting with that hitters play defense in real baseball).  Since everyone else is throwing 65% of their budget at hitters, if you look to build a good hitting staff with 50% of the budget, you'll never be able to afford players that can compete with the other teams.

Though I won't get into an old topic at great length, we did a study a few thousand leagues, thanks to data provided from CompuServe in the mid-1990s on standard Rotisserie leagues.  Of the 3-5% of the teams that spent less than 55% on hitting in these leagues, not a single team finished in the money in any league and typically, they finished approximately 10th out of an average league size of 14 teams.  Examining them more closely, we find that these teams did very well in the pitching categories, usually finishing near the top in three or four categories.  Looking at the hitting, they were almost invariably in last place, even though an "ROI analysis" showed that if they really were aiming for a good return on investment in the context of spending 50% on hitting, they were doing well.  Trust me.  Stick with 65%.  Given the number of readers we have, I know that there's a reader out there somewhere that has won their league using a 50/50 model.  Regardless, I don't see it as good basic fantasy baseball strategy and so my answer is: don't move to 50/50 ever.

This isn't to say that you can't "accidentally" spend 50% of your money on pitching but it should be because you valued players on the 65/35 model and just happened to end up with 50% of your money spent on pitchers, perhaps because all the good hitters were gone.  Even then, I'd suspect you would be in trouble but in terms of valuation, the 65/35 model works extremely well in most leagues.

From AskDL May 26, 2002:

Q.  I am curious how I should regard projections for the short term.  Should I play an inferior player, in terms of his projected total value, if I expect that for the short-term, he will be more valuable?

A.  This is an interesting question and readers probably face this quite frequently.  For example, you could have a player called up from the minors who will only play two weeks in the majors but is projected to steal 5 or 10 bases or something along those lines.  You know that during those two weeks, he's probably going to be more valuable than the other more valuable long-term player.

The best answer I can give is this:  If your league allows manipulation of the rosters in such a way that you can hold on to a "longer term" player but insert a "short term" player onto your roster, then by all means you should take advantage of that opportunity.  The problem is that I don't believe you could get away with this in most leagues.  Because of the limitations on taxi squads and benching players, most fantasy leagues will prevent you from getting such players onto your roster.  I certainly wouldn't be willing to give up a fantasy player with more projected seasonal value remaining.

Where you also face a challenge is estimating whether the player really is more valuable for the so-called short term.  You can't simply adjust fantasy values on a "per game" basis because part of the valuation of a forecast is the ability of a player to actually play games.  In other words, you might gamble that the so-called short term player is of value to you only to find out that he plays about two games a week.

Because it is impossible to convey the nature of a forecast to represent whether it represents a short-term burst of value or a long-term occupation of the bench, the only recommendation I can make is that if you must consider the short term in this fashion, you look at the forecasts on a "per game" basis.  In most cases, especially in the case of minor league callups, you will discover that we are forecasting the new player to get just 2 or 3 plate appearances per game, which should represent that we don't expect them to be regulars in the lineup, even for the short-term projected.

Too often, a player gets called up and the manager says "he's going to play every day" and then we get a series of emails from readers asking why we don't project this.  That's because in many cases, the manager doesn't yet realize that the guy is going to hit .150 over his first 40 at bats.  Those are players who end up back on the bench rather quickly.  For example, two weeks ago, when Homer Bush was released by the Blue Jays, we received several emails asking why we hadn't projected Joe Lawrence to be a regular in the lineup.  In fact, we had projected that Lawrence would be a regular for a few days and then wouldn't hit a lick and would end up back on the bench, bringing down his projected number of plate appearances per game.  The same is true of the so-called short term players.  Don't be fooled to believe that a player is a good short-term player because of his forecast.  If he has a good projection, chances are we will be projecting him to seize and win a full-time job, with rare exceptions, and to a limited extent, that will be factored into the overall remaining forecasted value.

If you can estimate short-term value by manipulating our forecasts and if your league allows it, by all means you should insert players with short-term value into your active lineup.  I can only emphasize that you shouldn't give up a player with greater total remaining seasonal value to accomplish such a short-term strategy.

Q.  Do multi-home run games factor differently into your projections than would a series of home runs in separate games?  For example does Cameron's 4 home run game tell you the same amount of information about his power potential as would a series of 4 home runs in 4 games?

A.  This is an excellent question and the answer is no.  We could easily fool ourselves by saying that a player has less of a chance to hit 4 home runs in a game if he is not a "real" power hitter but in fact, the science of statistics precludes you from using data that way for this reason:  In order for a sample to be meaningful in analyses such as we do, it must be, where possible, a randomly selected sample.  Now certainly, we are not using purely random samples in that we're more interested in recent data than older data but the reason for that is because we are trying to establish what a player's ability is now and the more recent the data, the better we can do that. In other words, if I were doing a political poll of potential voters, I would be more interested in data from this week as opposed to last week.  If we were to focus on a four home run game simply because a hitter hit four home runs, we would be committing a statistical error because now we are not dealing with a random sample but rather a specific game that is of interest to us.  A similar error could be committed by looking at a 20 strikeout performance by a pitcher.

In fact, let's consider this for a moment to prove an example:  Let's pretend for a moment that in a 4 home run game, a hitter gets 6 plate appearances.  It could be more and it most certainly could be less but we'll give him 6 plate appearances to demonstrate the example.  Now let's pretend that his real ability is 80 home runs per 600 plate appearances, which I think all readers would agree is beyond the reaches of any  player.  Now, this might surprise the reader but even our hypothetical 80 home run hitter should be expected to never hit more than 3 home runs in a game in his entire career!  Certainly, 3 home runs are within his expected range and he should achieve this mark at least several times in his career but 4 home runs has almost no chance to occur.

In that respect, were we to examine a 4 home run game on its own as proof of something extraordinary, we would be excluding some meaningful data (the player's other more recent games) in favor of data that interests or excites us.  Because we now are dealing with an individual game, and because most players will play many thousands of games in their career, we must use a level of confidence so high that it typically precludes us from excluding any data and thus, reveals nothing new about the player that we didn't already know.

There is a way, however, that a 4 home run game can impact the projections differently than, say, 4 home runs spread out over 4 games.  The way this occurs is because we do give greater weight to the more recent performance of a player.  Over the years we have developed and refined a complex system that assists us in providing weight to data from last week, last month, last year, etc... In that respect, a 4 home run game can reveal a bit more about a player in that it could be a sign of recent power as opposed to early power.  Obviously too, when a Mike Cameron or Shawn Green hits 4 home runs, it does impact our forecasts because they have instantly and dramatically increased their seasonal totals but were we to focus on the individual game, we would quickly discover that our level of confidence must be so high that we can't conclude anything new about the player.  To put it simply, no player in baseball, no matter how great, can reasonably be projected to ever have a 4 home run game in their career.  Were we to conclude something new about the player using that small sample, we would be wrong because using even a 99% or 99.99% level of confidence, we would incorrectly conclude that the 4 homer player is the greatest power hitter in the history of baseball.

Typically, our projections consider all seasonal data but we also do give greater weight to recent performance so in that respect, a 4 home run game could theoretically have a different impact on the forecasts than, say, a player who has 4 home runs in the first two months of the season.  But in general, a single game does not offer enough of a sample to make any meaningful conclusion though you can often and unfortunately forced to be evaluate players on such limited data.

From AskDL June 1, 2002:

Q.  I would like to know if you have an updated forecast for Chad Hermansen.  He looks to have the everyday job in right field now in Pittsburgh.

A.  Yes, and I still believe Hermansen is going to lose his job very quickly, as our latest forecast projects.  I see Hermansen as a .230s hitter with slight power at best and an incredible propensity for striking out.  I'll be very surprised if he's in the majors even a month from now.

As a late note, Hermansen strained his shoulder just before we went to press and underwent a cortisone shot this morning.  He's hitting just .242 this season in 19 games and I argue that even a healthy Hermansen won't stick.  While it may appear that he was getting "hot", he had just 11 hits in his last 55 at bats (a .200 average).

By the way, for readers who frequently ask, I should also emphasize here that there's never a need to ask whether any specific forecast is updated.  We consider each and every one of the players with every update and always evaluate their current role.  If we project them to get little playing time, compared to what the reader expects, we have a reason and usually the reason will be obvious, either because a player is injury prone or because we mention an injury in our player notes or more commonly because the player performance, as seen in the projection itself, is bad enough to cause a player to lose his job.

That answer also stands for those readers who write this column asking "Do you really think Player X will do this?"   We faced about a thousand such questions in the spring in relation to Morgan Ensberg and Carlos Pena and several recently in relation to Chad Hermansen.   If we say it, we really think it and there's no need to ask twice.  If we change our minds, I promise you'll know it by the new forecasts.

From AskDL June 16, 2002:

Q.  Has Keith Foulke lost the closer's role and if so, are there any Chicago pitchers worth picking up?

A.  It does appear that Foulke has lost the closer's role which is unfortunate because he's actually a better pitcher than his numbers make him appear to be and now, he may try to make an adjustment to win the role back.  In fact, that happens frequently with a player.  A good example is a hitter who goes into an 0 for 15 or 0 for 20 slump, which should be expected at some point throughout his career.  A hitting coach might tell him that he's doing something wrong when in fact he's doing everything right but just is experiencing the result of bad luck, the inevitable string of unfortunate outcomes that assembles itself on a player's statistical record.  Then, after the coach messes around with the player, he changes something that he's been doing and as a result, he's really in a rut because now, he's doing something new which might create even worse results in the long run.  The same is true of a pitcher who, in limited playing time, suffers from a few bad outings in a row.  A closer by nature is getting very little work and so even one bad game or two can make him look ineffective and thus, he loses his role, his confidence and then tries to change something.  In some cases, the damage can be permanent if he does tinker with his approach.

In fact, before leaving the topic of Foulke to discuss the options in the Chicago bullpen, Foulke had two problem games this season that skewed his results.  The first came on April 3rd in a game in Seattle.  The earlier you have a bad game, the worse it can make your season look because that game is part of your season-long record, constantly drawing your eventual performance in the wrong direction.  Because the April 3rd game came at the beginning of the season, this particular disaster made it appear that Foulke was pitching poorly for the entire season because that bad game was constantly included in any season-to-date numbers from the first week of the season.  In that game, which was just his second appearance of the season, Foulke allowed 6 hits and 4 earned runs in just two thirds of an inning as he was pegged with the loss.  Now, fast forward to the end of May, when Foulke was facing the New York Yankees, which like Seattle is another exceptionally tough opponent.  In that game, he didn't have his stuff and allowed 5 hits and 5 earned runs in just 1.1 innings to take his third loss of the year.  Now, while the nature of bad results must include single examples of worse games than normal, if you take away these two appearances, Foulke is 0-2 with 8 saves and a  2.45 ERA.  Removing those terrible two appearances, in 25.2 innings, he has allowed 22 hits, walked just 4 batters and struck out 21 batters.  That looks like the real Keith Foulke to me and I'd rather have a closer that does this for most of the season and then gets blown out in a couple of games.  I used the same example in reference to Billy Koch a week or two ago and the same is true of John Smoltz, who suffered one bad game early in the year.  Bobby Cox, of course, realizes what he's been getting because he doesn't overemphasize the statistical results that he knows include an unusual and uncharacteristic outcome.

The worst part about this whole analysis is that I do believe that the White Sox haven't really noticed this and because of that second appearance of the season, they have falsely believed that Foulke's having a bad year because his ERA started at 21.56 after game two and has been coming down ever since.  Thus, I do expect him to remain in a setup situation for a while and I sincerely hope, for his own sake, that he doesn't try to change anything to "fix" his so-called problems.

As for the rest of the bullpen, there are so many candidates for saves, especially Damaso Marte and Bob Howry.  I'm not fond of Marte's stuff and in fact, Howry's the only other arm in the White Sox' bullpen, other than Foulke, that I would have on my fantasy team.  Realistically, I'd be trying to trade for Keith Foulke if I had a chance because you'll never get him cheaper and he should be back as the closer by the end of the season, unless someone messes with his mechanics.  Buy low.  Sell high.

From AskDL August 3, 2002:

Q.  I notice that sometimes you seem slow to react to a change in a player's ability and other times you seem quick.  What is your reasoning behind adjusting a player's forecast?

A.  I know it sometimes frustrates readers, especially in April, when I don't dramatically alter a forecast to reflect a player's start.  Quite simply, April typically offers little information and a limited quantity of data from which to make a meaningful conclusion.  Yes, when a player gets off to a poor start or a great one, many readers want me to instantly alter a forecast.  Sometimes, a player can spend an entire season in a slump that is not actually representative of a decline or improvement in skills.  Subscribers of this site last year will recall that I spent the entire season continuing to forecast Johnny Damon to improve and it never happened.  The same is true this year for Moises Alou.  The belief with Damon last year was, quite simply, that he was a better hitter than he was showing.  This year, he started with a clean slate in Boston and lived up to such forecasts.

I should also emphasize that sometimes an April performance is so bad that it makes it appear that the player is having a bad year and that forecasts we published on May 1st did not come true.  In fact, many a May 1st forecast has gone exceptionally well but when mixed with the player's misrepresentative April, the player's season doesn't look up to par.

Basically, we hypothesize the ability of every player based on a combination of factors.  When the player performs outside of the margin of error, we make an adjustment.  When they do not, we might not make an adjustment but we also are constantly revisiting forecasts to see if there are new skills we didn't pick up.  A true .350 hitter in his first 100 at bats can hit anywhere between .208 and .392 and be performing within a margin of error of 95% confidence.  When he hits .208 over his first 100 at bats (or vice versa), so many readers want us to modify the forecasts and though we might, if we do so it will be because we have detected something new in his swing and not because we overreact to the trend.

I cannot overemphasize that many fantasy leagues are won on or around May 1st when a good deal of astute fantasy owners make deals to acquire players who are off to slow starts and trade those who are off to fast starts.  The mistake you must never make is what is called the "gambler's fallacy" in probability circles.  Don't ever believe that a player becomes "due" to do the opposite of what he has been doing.  If a true .350 hitter hits .210 in April, then he remains just as likely to hit .350 the rest of the way but once he hits .210 in April, his chances of finishing with a .350 average for the season have diminished.  In other words, don't make the mistake of concluding that if a player starts slow that his season will somehow "balance out" and that he will be due for a slump later.  When assessing a player, we project what he will do for the remainder of the year and it's often a sore spot among readers who want to see what they expect.  We'd rather be right than make everyone temporarily happy.

Q.  If you had to identify the biggest weakness in fantasy baseball, in terms of how accurately it represents real baseball, what would it be?  The way defense is or isn't handled?

A.  While the lack of most leagues accounting for fielding ability is a major oversight, it's actually not the biggest hole I see in FBB.  I would say that the inability to force your prospects or veterans to play is the biggest problem.  So many players lose their jobs in April and May even though they are actually better than their statistics imply.  If you really owned or managed a baseball team, you could keep them in the lineup or rotation as long as you desired and sometimes, you would ultimately be proven to be right about a player.  I see this as the hole that can't be fixed in fantasy baseball because it is attached to how players actually perform.  There are many ways to consider defense but it's almost impossible to account for players whose real teams give up on them before you would.

One solution to this, albeit a complex one, would be to award secondary points for Triple-A performance.  In other words, make a major league home run worth 1 home run and make a minor league home run worth 0.5 home runs or something along those lines.  It wouldn't solve everything but at least it would introduce an element of opportunity into the game where minor league players get some recognition for the skills they have and you correctly get rewarded for their major league readiness, despite what their teams think of them.  Of course, this creates scoring nightmares and most fantasy scoring services, if not all, couldn't handle it.  I throw out the suggestion as an option to solve what I consider to be one of the biggest flaws in fantasy baseball.

From AskDL August 11, 2002:

Q.  I have been having a heated debate with a friend over the future production of Eric Hinske vs. Aaron Boone.  His point is that since Hinske is only a rookie, he'll continue to get better while Boone is having a fluke year.  I argue that a player that has a break out year later in a career is more likely to maintain that level than a rookie is.  I use the examples of Luis Gonzalez and Sammy Sosa to prove my point while the examples of former rookies of the year flopping later are too numerous to mention.  Could you please give us your opinion on the matter and any facts you have to support it?

A.  Well, I hate to say it but I think you're both attempting to make a generalization based on specific examples you can recall.  Certainly, Luis Gonzalez and Sammy Sosa did eventually have breakout years but for every player like that, you can also find a Brady Anderson who had a so-called breakout year with 50 home runs and never returned to that level.

Similarly, while there are many rookies who seem to flop in their sophomore year, part of that is because in general, we notice the Rookies of the Year who don't repeat extraordinary seasons and great seasons typically do not repeat themselves because they are usually on the upper end of likelihood.  A Rookie of the Year winner, as winning the award probably tells us, may have performed on the upper end of his ability but there are many such players who also go on to have even better careers than their rookie seasons and you shouldn't be quick to ignore those.  Consider that Scott Rolen was the Rookie of the Year in 1997 and improved in 1998.  Mike Piazza won the award in 1993 and went on to have better seasons later in his career.  The same is true for many other Rookies of the Year such as Eric Karros, Nomar Garciaparra, Derek Jeter, Tim Salmon, Chuck Knoblauch, Jeff Bagwell, Mark McGwire, Jose Canseco, Darryl Strawberry, Cal Ripken Jr. and Fernando Valenzuela and these are just the Rookies of the Year that come to mind as I consider your argument about the number of rookies that "flop."

My belief is that whether a player made his debut in the majors this season or is having a so-called breakout year, the chances that he may be overperforming or underperforming are relatively the same.  The nature of statistics tells us that some players will perform above their ability over a certain period but we also know that a player can actually improve and become something he wasn't.

Since you've asked my opinion on the matter, I believe that Bret Boone did perform above his actual skill level last season and that Eric Hinske is performing slightly above his skill level this year.  If you are asking whether I think either of you is correct on the rookie vs. breakout argument, my answer is that both would be hasty generalizations.  To make a proper assessment of a player, you need to scout him and you cannot rely purely on statistics.  Hinske is showing better bat speed than he showed in the minor leagues last year but not enough to explain his great performance.  Sammy Sosa and Luis Gonzalez showed markedly better bat speed in their breakout seasons, which is why we were able to safely project them to maintain their newfound pace.  Lance Berkman this year, for example, has only shown slightly improved bat speed which is why I have not jumped aboard the Berkman bandwagon and is why I am currently likely to project a dropoff season in 2003, even though he should still have another great year.

I wish I could give you a better answer but the best way to find out who is right about Eric Hinske is to see what kind of career he has and even at that, it will only support the argument in relation to Hinske and not necessarily rookies in general.

Q.  I wish you would write more about fantasy baseball strategy.  Why don't you?

A.  I think we do offer plenty of fantasy baseball strategy but the main focus of what we do is forecasting baseball performance.  Now, clearly the ability to forecast performance is the most important aspect of fantasy baseball success and while I do tend to discuss fantasy baseball quite a bit, I don't like to think of it as the exclusive target audience of Baseball Notebook.

Many fantasy leaguers should remind themselves to stay focused on "real" baseball because fantasy baseball success will come from a better understanding of the dynamics of real baseball. For example, if we discuss a topic such as the ability of a player to produce "real" runs in baseball, that is of key importance because if a player can produce runs, he will increase his chances of sticking around in baseball and that directly will determine whether he becomes a long-term producer of home runs or RBI or whatever statistical category interests the fantasy baseball reader.

Over the years, I've strived for balance in these pages and I do believe that we've made great strides in adapting our information for both a fantasy baseball community and a strong base of readers whose primary interest is the real game itself.  We will continue to focus on our primary mission of constantly improving the accuracy of baseball forecasts and we fully believe that fantasy baseball success, more than anything else, benefits from our concentration in this area.

From AskDL October 14, 2002:

Q.  What do you think went wrong with Keith Foulke this past season?

A.  Absolutely nothing.  He was the victim of a few bad appearances and an organization who became convinced that he wasn't as good as he really was despite all the elements of his game looking the same.  His pitches were all there and his command was as it always has been.  In fact, despite the lack of saves, his season was typical of previous seasons.  He finished with a 2.91 ERA, better than his 2000 season and under 3.00 for the fourth consecutive year.  He walked just 13 batters in 77.2 innings, a better ratio than a year ago and with two intentional walks thrown in.  As I said back in mid-summer, he was undeservingly dropped from the closer's role and I hope that the White Sox realize that he deserves either his closer's role back or a chance to start in 2003.

From AskDL February 9, 2003:

Q.  If saves are only 10% of the league points available (in a 5X5 league) in a league that has a $260 salary cap, shouldn't most of the best closers be in the $26 range (10% of $260)?

A.  No, because of the problem with unbalanced bidding.  That is, traditionally leagues only spend about 35% of the money available on a pitching staff.  That means that if if you have a 25-team league spending $260 each, there will be a total of about $2275 (which is 35% of the total money spent in the draft) spent on pitching and if there are five categories, if we were to weigh them equally between the other categories, that means that to win, the total amount spent on saves should be about one fifth of this amount or about $455 in the entire league.

Now, if that $455 figure were correct, that means that each pitcher that will be drafted has to be considered in terms of what he brings, percentage-wise, in terms of saves.  Looking at 2002, let's pretend that every pitcher who was destined to end up with at least 10 saves were going to be drafted in the auction.  The total saves that would have been available in 2002, using that cutoff point (and the real cutoff point is actually even lower) would be a grand total of 1044 saves.  That is, adding all the saves up for pitchers with at least 10 saves last year gets us 1044 saves.  That means that a major league leader, such as John Smoltz with his 55 saves last year, is contributing 5.27% of the saves that are likely to be drafted.  If the league has $455 to spend on saves, 5.27% of that is just $24 in terms of Smoltz's contribution to the saves category.  While he may bring additional contribution in terms of his ERA and/or WHIP, it is not as significant as you might think because there aren't enough innings for him to make a huge difference unless he gets hammered (I'll address this in a moment).  Remember too that in the projection set, you don't really see 2002 John Smoltz like seasons because it's not good forecasting to be projecting on the high end of expectations.

Now, I will concede two things here.  Firstly, whether saves should be evenly weighted in terms of the pitching categories is open to debate though I believe if it is our goal to win the league, we must weigh the categories equally and I am not a believer in the theory that we should weigh more to spending on saves and less to, say, ERA.  I know it is not a popular idea that I argue that closers are overvalued in most leagues and I don't necessarily advocate tanking a category deliberately.  At the same time, you're trying to win and not trying to do what's expected of you.  I have contemplated that in our interactive forms we should consider functions that would allow the user to create "unequal weights" to the categories so that within pitching, you could decide that saves gets a higher percentage of dollars than, say, wins.  I think it would be poor strategy but I suspect there are users who would use it and so I will look into the technical considerations of having an interactive form such as this.

The second concession is as already mentioned, that there are pitchers who do contribute in the other categories beyond saves and I must emphasize that our ranking forms already consider that when they produce projected values.  I do want to give an example here though that may surprise you.  Let's consider two pitchers from 2002, one in Derek Lowe and the other in Kazuhiro Sasaki, certainly one of each of the elites from 2002.  Now,  let's say you have an average major league fantasy squad made up of eight pitchers from last year and you're going to add one more to make a nine man pitching staff.  Obviously, the smaller your fantasy league, the higher the bar is set and the better the overall stats but let's go with an average fantasy team from a 25-team league, which might be on the low end of the fantasy spectrum for 14-team and 12-team leagues.  We're only going to look at ERA for this exercise.  Here's our hypothetical team before we decide to add the 2002 Lowe or the 2002 Sasaki:

Innings Pitched: 1254
Earned Runs Allowed: 573 (team ERA is 4.11)

Now, we are trying to compare what sort of difference a pitcher like Johnson makes when compared to a Sasaki, in terms of ERA, especially considering each has about the same ERA as the other but a different number of innings.  Let's see what our team ERA looks like with either added:

If we add Sasaki, we now have:

Innings Pitched: 1314.2
Earned Runs Allowed: 590 (team ERA is now 4.04 with Sasaki added)

If we add Lowe instead of Sasaki we end up with:

Innings Pitched: 1473.2
Earned Runs Allowed: 636 (team ERA is now 3.88 with Lowe added)

The example's not intended to apply to every fantasy league but at least consider the example.  A Kaz Sasaki with his 2.54 ERA only lowers our team ERA by about 7/100ths of an earned run whereas a Derek Lowe, with a similar 2.59 ERA but substantially more innings, has a huge impact on our team, improving it almost a quarter of an earned run.

The point of all of this is to emphasize something:  We go into fantasy leagues with a pre-conceived notion about what player values should be because traditionally, and in most of the published materials, they claim that relievers are worth quite a bit.  While I do concede that they have an important role to play and while I don't advocate deliberately dumping any category, you also can't get around the math that typically, they don't make the difference to winning that most league salaries would justify and I believe that the day is inevitable when there will be a paradigm shift in fantasy baseball, one that will result in players stocking up on great starting pitchers, dominating the wins, strikeouts, ERA and WHIP categories and changing the overall league opinion about the value of closers.

Again, while I don't recommend deliberately deciding not to have any saves, consider that if you did go with nine starting pitchers, you would almost certainly finish first in the wins and strikeout categories and would be a strong contender for first in the ERA and WHIP categories.  In a 10-team 5X5 league, with a maximum of 50 pitching points available, you're on your way to somewhere between 30-40 points.  Mix that with a strong hitting staff and that makes you a solid contender, even if you end up without a single closer.

From AskDL February 23, 2003:

Q.  What stage in a fantasy draft is the best for getting bargains?

A.  It happens throughout the draft but you can really clean up when opposing rosters don't have any slots free for players you want and most typically, this happens when your opponents have already filled all their first base and utility spots.  At that point, you can clean up by drafting first basemen ultra-cheap and I remind readers that you should always calculate what opponents can actually do before calling out a name and/or bid value.

The reason for this is twofold.  Firstly, if you don't have an opponent who can even take a first baseman, then in a straight draft, there's no good reason to be calling out any first basemen names because no one can draft them anyway and so you should wait.  In that case, wait until the end of the draft to get you first baseman and instead, go for other players in the meantime.  Better yet, in an auction league, if your opponents don't have any room (or money), you can often clean up with a $1 first baseman.

Most frequently, the bargains come about 70% of the way through the completion of the draft and they are most common in the first base slots and on rarer occasions, third base and outfield slots.  Pitching tends to remain at a premium throughout the draft.

Q.  I understand your strategy about a 78% maximum bid but the highest priced player on my projected values is $30 and most are under $20.  If I follow this strategy, I will never get any of the superstars.  With a great keeper list, I'm afraid I'm going to have money left over after the draft.  Do I really not need the superstars to win the league?

A.  One of the realities of auction drafts in fantasy baseball is the presence of something called "draft inflation" where there is more money to spend and not enough players.  Technically, if you wanted to ensure that you didn't have any money left over and you were correctly bidding, you would have to be monitoring for the stage in the draft where inflation kicks in.

My recommendations re: the 78% maximum bid attempts to address this indirectly while allowing you to focus on your own roster.  You see, once you get your first few bargains, you will be able to begin overbidding for remaining players because your so-called "bid modifier" will change during the draft.  That is, if you clean up by grabbing your first $1 player who might be worth $15, and believe me you will find one of these, suddenly you're in a position to start overpaying for players.

What you need to do, as I've mentioned in previous columns, is keep two running totals throughout the draft.  The first column is your needed projected value column, which we are trying to have total $333 at the end of the draft (in a $260 league - adjust for your league accordingly).  The second column is the amount you have left to spend, which before any keepers are announced is $260.

The first column is constantly being reduced to reflect the projected value of players you acquire so your initial $333 amount should be reduced by the projected value of your keepers.  The second column, which starts at $260, should be reduced by the amount you've spent on keepers.

Now, throughout the draft, keep reducing both columns and keep a calculator handy.  As soon as you've reduced both columns for the latest player you've acquired (do this immediately so you'll always have the next calculation already done), calculate what column B is divided by column A and that's your current bid modifier at any point of the draft.  So, if you still need to acquire $250 in player value and you have $220 left to spend, your current bid modifier is $220 / $250 = 88%.

That means that when a player's name comes up and you want him, you take his projected value and multiply it by the current bid modifier and make your bid accordingly.  Of course, when you have the chance to make the final bid in a round, never bid more than you have to but at least this sets the ceiling for you.

What will happen is that around the mid-point of the draft, if you have a strong keeper list and players are going in typical fashion, your bid modifier will move above 100% and you will now be bidding full value.  It usually takes me until about two thirds of the way through the draft that I have very high bid modifiers and now I can afford to deliberately overpay for players to round out the roster.

Yes, early in the draft it will seem like you are getting few players but when you get the first bargain or two and if your keeper list is decent, you will start to see this phenomenon kicking in.  If you have money left over, chances are good that your keeper list wasn't good or your opponents have virtually identical opinions about what the correct bid amounts are for players, in which case you will have a problem.  That's rare and if you go in with this sort of a plan, you will be able to bid extraordinary amounts late and get precisely the players you want.

In terms of superstars, you can usually only afford to get them if you have great keeper lists and believe me, you don't need A-Rod or Randy Johnson to win.  They help but there's a whole second tier of great players, some of whom will end up on your roster if you follow such a bidding strategy.

Q.  When working with your ranking forms at the website, I believe you advocate a 65/35% split between hitting and pitching.  It seems to me that a consideration of the roster and categories is more accurate.  For example, if 13 batters and 9 pitchers are being chosen, 41% of the players chosen are pitchers and 59% of the players chosen are batters.  Shouldn't I use that sort of split instead?

A. No, because fantasy leaguers don't think that way.  For whatever reason, in all the leagues we've analyzed, regardless of the make-up of roster requirements and splits between pitching and hitting, fantasy leaguers tend to think in terms of weighting hitters at around 65% and rarely do I see this fall below 60% or go above 70%.  In fact, in a study of over 400 leagues of real data in a variety of leagues, we found just a single league that went as low as spending 59% on hitting.  Trust me.  It's not that pitching isn't worth as much but it's because if leagues tend to think that way, then you need to be thinking the same way too.  If a league views that it has 65% of its money to spend on hitting, then you need to be thinking that you have approximately two budgets, one a hitting budget and one a pitching budget.  I've experimented heavily with modifying the split and even have tried the 50/50 split between weighing hitting and pitching equally and what ends up happening is you get slammed in the hitting categories and do only slightly better in the pitching categories.

From AskDL March 2, 2003:

Q.  Is there a way to adjust for keepers on your fantasy ranking forms?  If A-Rod, Nomar and Jeter are all on keeper lists, shouldn't all the dollar values for shortstops be adjusted up or down to reflect the frozen dollar amounts?

A.  This is a fantasy strategy I don't believe in as previous columns at the main website have outlined.  I believe that you should think of projected values in terms of the overall value of the league and focus specifically on building a fantasy roster that has a desired total value (I usually advocate $333 for a standard $260 salary team in most 10 or 12 team leagues) with a certain amount of money.

While I don't wish to repeat myself in terms of other essays posted at the website, the projected value of a player does not change if he is kept by a team and what that means is that the player's remaining in the pool don't see a change in value either.  Certainly, the return on investment varies and obviously, the strength of your own keeper list must affect your bidding strategy but let's say that we increase the values of remaining shortstops because of your example.  Now, we fast forward to immediately following the draft.  Is Chris Woodward's value inflated relative to A-Rod if we bumped up Woodward because so many good shortstops were retained?  A home run remains a home run and an RBI is an RBI.

I have always maintained that you should modify your bid amounts based on your needs and your remaining resources (i.e. money left) but do not make the mistake of focusing on keeper lists of other teams more than you need to.  It's a consideration but you have one goal and that is to put together your own winning roster.  Once you determine the amount of value you need on your roster and how much you have to spend on it, you can focus on that.

I know there are readers who disagree with this and I've heard the arguments for the differences.  In fact, there really are two schools of thought on how to account for keeper lists and I'm in one group and I know many readers are in the other.  Every reader must form their own opinion but if you're asking my advice on how to build a winning team, you have it here.  Don't adjust the projected values based on who is kept by your league.  Adjust your bidding strategy based on how good your own keeper list is and you'll do quite well.

For those who do believe in adjusting based on the quality of keeper lists, then by the same token, you will need draft software to get you through the draft.  By the same reasoning that we would adjust remaining values based on who makes it on to keeper lists, we must also adjust based on every undervalued or overvalued player that is sold during the draft.  Those who believe in such things are the types who would use draft software and under such a belief, they would need it.  I'm one who says you don't need it and admittedly, I've changed my opinion much in this regard since ten years ago.

Q.  Using your strategies, I've traded many of my overpriced players and have assembled a team of 13-15 players of $5-$10 value that are worth a little more than their price tag.  I don't have any top offensive players.  Should I overbid to get them?

A.  When you say that they're worth a little more than their value, I do hope that the sum of your protected salary is no more than about 75-80% of the projected value of the players.  If you've done better than this, you can automatically afford to overpay for superstars.  If not, maybe you don't want to (or shouldn't have already) protected so many players because you still want to leave room to get that target amount of value.  Since there's no such thing as a $100 value player, at least in most leagues, you must leave room for the desired value on your roster.

So, though I don't recommend overbidding, I do believe that if you've really assembled a budget keeper list that produces superior value, you will already be in a position to overbid for superstars and no one else in your league will be able to afford to match you on many of the players you need to round out your roster.

Q.  Using your projection sheets in my draft league, I am concerned that I will pick a player poised for a breakout season much earlier than the typical owner in my league.  Do you have any strategies that would prevent me from selecting a player "too early?"

A.  Yes.  First, look at the relative confidence rating for players and in your first pass, try to skip over the lowest confidence ratings.  For example, a Danny Graves or Byung-Hyun Kim have excellent projections but both have extremely low confidence ratings.  This would have been true for our projection of Derek Lowe last year, whom we projected to have a great season but I know had a lower degree of confidence then, even though we didn't publish confidence ratings last year.

This is a clear case where draft leagues are different than auction leagues because in an auction league, you can throw out the sleeper name and usually get him for a low salary.  In a draft league, you are always trying to ask yourself these two questions and I want you to mentally take these with you to the draft:  (A) If I skip over the highest remaining name on my list, is there a fairly certain chance that he will still be available by the time I get to next choose a player?  (B) If I think he will still be available by the next round, will I still have room for him by taking another player instead?

The reason for the first question is obvious.  If no one's going to take him, then don't waste the early pick.  The second question is asked because many fantasy leaguers think a player will still be in the pool so they move down their list to take a different player.  When it comes around to them, they realize that they just plugged the only hole they had for the sleeper with the player they previously took.  Don't make that mistake.

From AskDL March 9, 2003:

Q.  In my league, I am sure that on average, the league spends about 68% on hitting instead of 65% on hitting.  If I am positive of this, should I change the ratios in my projected value ranking sheets?

A.  Yes.  If you're positive that your league spends a certain ratio, then you can and should think in those terms as opposed to the 65/35 split.  In most leagues, 65/35 works exceptionally well but in those few leagues where you have an owner who constantly loads up on hitting or pitching year after year, disregarding the other side of his roster, then you should bring this information into your decision on how to weigh hitting and pitching.

Q.  I am in a straight draft league with no bidding.  I am wondering how I can tailor your ranking sheets to my benefit.

A.  Use the dollar values as an estimate of relative value and as I've said previously, in cases where you think a player will last until a later round (i.e. a sleeper), bypass him until you absolutely have to take him.  It really isn't difficult and many people are in draft leagues as opposed to bidding leagues.  Just regard the projected value as an estimate of projected relative value as opposed to any sort of advice on what to bid.

From AskDL March 23, 2003:

Q.  In your top 100 pitching prospects, I'm amazed that Jesse Foppert and Gavin Floyd did not make your list.  Please explain!

A.  Without question, this was the most common question in emails to this column this past week and it's always difficult to answer the question "Why don't you like this guy?" when in fact, I can think that either or both of these pitchers (or any of the other 10-20 pitchers other emails mentioned) may end up as good or even great major league pitchers.

Let me first emphasize that if you have a top 100 pitching prospect list, a list in which the names on it will peak at different times, what you are really endeavoring to do is to create a list of who will be the top 20-30 pitchers in baseball, at different points in time.  In other words, if Pitcher A is a top pitcher in 2005 and Pitcher B is a top pitcher in  2009, they both warrant mention on a top 100 list but it isn't to say that both will peak in the same year.  Therefore, if we're talking about players who will one day be in the top 20-30 of pitchers, at some point of their career, what we're really trying to do is give you a list of pitchers who will one day be within the top 3-5% of the pitching population when they do in fact hit their prime.  That's a very precise segment of the forecasted prospect community.

With that in mind, my answer quite simply is that I don't currently believe that either Foppert or Floyd (or any of the others who weren't on the list but met the list's criteria) will one day be among the top 5% of pitchers and all of the other players do meet this expectation.  It's tough to leave out a pitcher when everyone else, including established insiders in baseball, expect great things from a player.

I should first speak to many of the readers who wrote things such as "publication X ranks him as the top prospect" or something along those lines.  To that segment of the reader base, I must pose the question:  Have you actually seen him pitch, either in person or on video?  If not, then what you're asking me is this question:  "Why didn't you include him?  Scout X says he's good or Publication Y says he's great."  I can't explain other than to say that my assessment of his skills varies from these people.  When I watch Foppert pitch, despite how his minor league results look, I'm not as impressed as everyone else.  Yes, he's a big guy with obvious major league stuff but I'm not trying to project whether he's going to be a major leaguer.  I'm trying to project if he's going to be a top major league pitcher.  I am aware that the deservingly well-respected publication Baseball America has named him the Pacific Coast League's best prospect and to be honest, I don't see what they see.  If I'm wrong, well them I'm wrong but if you disagree with me, at least do so because you've seen him pitch and believe that he's better than I think he is or because you trust these other publications more than our findings, which you are obviously entitled to do.

I face these sort of questions each and every year, dating back to when I published my first top 100 prospect list in 1994 in pre-Internet days, when Baseball Notebook was a single numbered CompuServe email address.  Since we published that first prospect list, we have left off many highly-touted pitchers, not because we like to be the contrarian but because we didn't think they would have great careers.  Over the years, each time we do this, I've personally faced a plethora of letters and emails saying "this guy is great" or "you must not know anything to leave that guy off."  Among players who never cracked our top 100 at any time and who generated plenty of angry responses have been Paul Wilson (you can decide for yourself if you think he ever became a top pitcher), Kris Benson (ditto) and Matt White and I think Kris Benson in particular was the one who met the most disagreement from readers who said it was "obvious" that he was going to be a great pitcher.  Maybe he will still be a great pitcher and in fact, I expect him to finish his career much stronger than he started, but I think that my expectation about his overall career has proven to be more accurate than had I listed him among the top prospects.

Returning to the idea of Foppert not making the list, his stuff reminds me of a young Erik Hanson and he's a bit older than you might expect too (22).  When a pitcher's stuff and potential upside don't impress me, especially when others say he's going to be a great pitcher, I investigate to see if the statistics offer a different perspective.  His 2002 minor league translation (and these translations usually do a great job of accounting for the level of competition, the league and the park) is a 5.52 ERA.  In 132 translated innings, he allowed 139 translated hits, 74 walks and struck out 143 (this to his credit).  Looking at his performance without translating it, just in case you don't believe in our minor league translations, Foppert had a 3.98 ERA in 79 innings spanning 14 starts with Triple-A Fresno.  On the same Fresno team, there were three pitchers who had more innings and a better ERA.  Kurt Ainsworth, about two years older than Foppert, pitched 116 innings with a 3.41 ERA.  Jerome Williams, who made our top 100 list is actually about a year and a half younger than Foppert and posted a superior 3.59 ERA in 160.2 innings for the same Triple-A team and with a much better WHIP (1.183) than Foppert (1.342).  Granted, Foppert's strikeout to walk ratio was better than Williams but I don't put the same emphasis on strikeout to walk ratio as I do other statistics, such as the ones mentioned here.  If I put much weight on strikeout to walk ratio, then I'd have to change my mind about Rick Reed, Eric Milton, Brian Lawrence, etc...  The short version of what I am trying to say is simple:  When both my own scouting of a player's abilities and my analysis of his statistics says that he doesn't deserve to be listed among the "top 100 prospects," I will not include him simply because the majority of people say he's a top prospect because then, I don't have a reason to put him on the list, other than because of what everyone else says.  At that point, I would no longer be offering my own insight but rather deliberately plagiarizing the opinions of the others, which I will not do.  As I say, Foppert reminds me of Erik Hanson and he could very well have an Erik Hanson type career.  That wouldn't put him among my top choices.  If I'm wrong, so be it.

Just to briefly speak to the Gavin Floyd part of the question, he did have a great year at Single-A Lakewood and I have seen him pitch much less than I have seen Foppert.  My initial perception of Floyd's stuff is that his skills remain quite rough, which is to be expected for his age (20).  As for his statistical results, which seemed extraordinary, we must remember that the ball doesn't carry in the Lakewood park in which he spent half of his time and more importantly, the South Atlantic League is filled with pitcher's parks.  In fact, the league was the second-best for pitchers in terms of average number of hits allowed per inning and the SAL was easily the lowest in terms of home runs allowed per inning by pitchers.  Our converted season for Floyd's 2002 year is a disastrous 6.15 ERA in 152 translated innings.  He may end up as a great pitcher but again, neither my scouting of his stuff (albeit more limited than Foppert's) nor my analysis of his statistical results warrants putting him on the list.

If you don't agree with either, then obviously you should insert your own opinions.  I only caution readers that when inserting your opinion, you will want, as much as possible, to decide whether you are doing so because of the majority opinion about a player or because of your own scouting.  While either could be a good reason, at least if you have a clear recognition of why you are ranking a player higher than someone else does, you'll learn in the future whether you need to refine your own scouting methods.

If you disagree with my methods or that I omit players from the list, well then at least you can be certain that I always say exactly who I think will be good or bad, regardless of how I think a list will go over with readers or what anyone else is saying.  This is not to say that I ignore other opinions and in fact the opinions of top publications usually motivate me to further investigate what I am missing about a player.  I was called more names than I recall for leaving Kris Benson off my top prospect lists years ago and the one thing I always say to readers who disagree with a projection is simple:  The future remains unwritten and I might actually be right.   If I'm wrong about a player (as I have been about the direction of several prominent current players' careers, including Tony Batista and Orlando Hernandez, among others), I'll be the first to tell you.

Q.  Of current major league hitters, who do you feel will make the jump to perennial All Star numbers over the course of the next five years or so?  My feeling is that Burrell, Beltran, Berkman and even Dunn are a year or two away from being annual MVP candidates.  What do you think of this and can you give me an idea of where these players will max out in a given season?

A.  This is a great question and it appears that you are asking about established players (meaning that for the Josh Phelps and Hee Seop Chois of the world, I refer you to our recently published top prospect lists).  I should note that a player's career year is almost always above their real ability.  That is to say that when a player has that year where they hit .370 and they never return to it, it's almost always because they performed on the high end of their abilities, still within an expected range and certainly worthy of praise but on the high end of the range.

I'll start by answering the first part of your question and then deal with the specific players you mentioned.  Without question, and our current projection reflects this for even the upcoming season, I believe that Albert Pujols is on the verge of becoming one of the best hitters in all of baseball, if he isn't already and well beyond what he's already accomplished.  Our current projection for 2003 is that he will hit .334 with 38 home runs and 129 RBI, the average and home runs of which would be career highs.  I believe that within the next five years, Pujols will graduate into being a consistent 50 home run a year player and it's within his ability to eventually have a 60 home run season before he's retired, at least in this era.

After Pujols, you rightfully included Adam Dunn in your list and I expect that within three years, Dunn will be a consistent 40+ home run hitter, capable of a 50 home run season in a good year.  I still believe Eric Chavez hasn't even come close to his real ability (so far his career-best year of 2002 saw him hit 34 home runs) and he's going to hit 45-50 home runs in a season before he's done.  Ditto for Andruw Jones, who has thus far peaked at 36 home runs but is likely to continue his development into a 40-50 home run per year hitter.  One guy who will be just below this group is Corey Patterson, if he can learn to take a walk.  He is going to become a reliable 30 homer guy and interestingly, I used to rate him lower and by the time everyone else had soured on him, I had changed my mind.  He has an explosive bat and underrated speed.

As for the players you mentioned other than Dunn, I see great things but I don't believe they'll be consistently in the MVP class, even though all are capable of having individual great seasons.  I actually believe that Beltran has already been performing on the upper end of his abilities and I don't expect him to become the 40 home run threat that some do.  Pat Burrell is much older than most people think (26) and so he should already be in or close to his prime.  I actually think that we shouldn't expect too many seasons like 2002 from here on.  I believe that we just witnessed Lance Berkman's career season and I'm expecting 30-35 home runs a year, on average, until he reaches his declining years.

From AskDL April 6, 2003:

Q.  I've read recently about the "law of averages" and someone suggested that if a veteran hitter hits .150 in April, that means he's likely to hit better later in the year and it balances out.  Do you agree with this?

A.  No, I don't and neither does the law of averages.  What the law of averages says is that the larger the sample, the more likely an average result will be closer to the real ability of a player or coin flip or whatever you're testing.  It does mean that if the player is better than a .150 hitter, say a .300 hitter, that the larger the sample size, the closer you'll see a result that reflects his real ability.  Unfortunately, the law of averages only refers to what happens from now on.  In other words, if a player hits .150, the law of averages does NOT mean that he should balance out by hitting .450 somewhere else.  That .150 start is mixed in with everything he does from this point on and while players, unlike coin flips, do have memories, you'll often find that a slow starting player is the same great player that you saw a year earlier.  Since April results get mixed in, if I really believed he was a .300 hitter, I would forecast him to hit .300 the rest of the way and that means he would finish with a sub-.300 average when you mix in the .150 average.  The law of averages doesn't compensate for the terrible start.

What you must keep in mind is that even if the law of averages doesn't mean he'll finish with a .300 average, if he really is a .300 hitter he should be forecasted to hit about .300 the rest of the way and we should only be interested in what a player will do the rest of the way.  So, you must examine his ability from the perspective of what you believe has not already been put in the books and can actually help you.  Therefore, that's why astute readers will notice that as we update our projections weekly, we always only list a projected remaining forecast and not a projected final total for what a player will accumulate.  What he's done is history and does you and your fantasy team no good.

Q.  Why do you project such a drop in steals for Fernando Vina, from 17 last year to just 10 this season?

A.  If you look at the latest caught stealing projection, it has Vina getting caught 6 times the rest of the way, succeeding 10 times.  Simply, my belief is that he'll run almost as much as last year but will get caught more often.  That in turn should naturally lead to a slight corresponding reduction in attempts as it becomes clear that he's only succeeding 63% of the time and thus isn't doing the team any good.  Last year, his 17 steals came with a 61% success rate.

Your question enables me to jump the gun on a future essay topic I had been considering and it's an aspect of forecasting I haven't really talked about before.  The idea simply is that you will be amazed when you look at what happens to a big-time base-stealer who doesn't succeed at least two thirds of the time.

Since 1969, there have been a grand total of 3,108 players who were coming off of seasons in which they stole at least 10 bases and who returned to play the following year.  Let's split those players into two groups.  The first group is made up of players who succeeded at least 66.7% of the time (i.e. two thirds) and the second group is made up of players who didn't.  The successful group has a total of 2,339 players and the "unsuccessful" group has a total of 769 players.  Of the 2,339 returning players who stole at least 10 bases and were successful two thirds of the time in year one, 67% of those (1,558 to be exact) went on to steal at least 10 bases again in year two.  From the second group, just 55% of the group (426 of the 769) went on to steal 10 bases in year two and the reason for this was obvious - their managers picked up on their inability to steal, often a newfound problem as a result of age, and stopped giving them the green light.

So, in Vina's case, I expect his propensity for getting caught will stop him from getting as many green lights.

From AskDL April 27, 2003:

Q.  Do minor league statistics impact your updated major league projections?  In other words, if a projected September callup is tearing up the minors, does his major league projection change as a result of this?

A.  Yes, minor league statistics are always considered in projection updates, for any player projected to eventually appear this season.  If a player is dominating the minors, we have to evaluate whether his performance represents a real change in abilities but for every player, we make such an evaluation.  A great example of this is that we downgraded Rick Ankiel substantially in the latest forecast update, largely because of his terrible performance at Double-A this season.

From AskDL May 18, 2003

Q.  I have noticed recently that there are a number of pitchers that are winning with terrible K/BB ratios and with ERAs and WHIPs that don't sink their owners' fantasy teams.  Why is this so?  I usually insist on at least a 2/1 K/BB ratio to convince me that a pitcher is in command of his game and should either break out or continue along his current road to a successful season.  Your thoughts?

A.  Most readers who have read my material know that I've never been a huge fan of the admittedly popular strikeout to walk ratio, even though I recognize that it is unquestionably linked, to a degree, to ultimate performance.  As I've said before, if strikeout to walk ratio was really the key indicator of a pitcher's ability, pitchers like Bob Tewksbury should have had Hall of Fame careers.

I do recognize that it does tell us something and certainly, if we were to separate pitchers into blocks of pitchers, K/BB ratio is a rough indicator of superiority.  However, I recommend against setting any criteria that limits you based on K/BB ratio alone.  It's simply another factor we have to consider and if we see it that way, we'll be better prepared to accept and understand that some pitchers can have terrible K/BB ratios and be great pitchers and others can be great K/BB ratios and still not be Cy Young candidates.

A great example is Kirk Reuter.  Here's a pitcher who finished with a 3.23 ERA last year and has had an ERA under 4.00 six times in his career out of ten seasons, albeit some of those seasons only having as many innings as a relief pitcher would pitch.  His strikeout to walk ratio last year was a dismal 1.41, in the bottom third percentile of all pitchers in the majors last year.  In fact, he's doing it again this year with more walks than strikeouts but a fantastic 3.00 ERA.  Rueter is a pitcher whom I project to continue having success, with a forecasted ERA under 4.00 in the latest forecast set I have published.

Conversely, we could consider Bobby J. Jones of the 2002 Padres.  Last year, his strikeout to walk ratio was 2.86 and he was better than 85% of the pitching population in strikeout to walk ratio.  His results last year?  A 5.50 ERA, a 1.435 WHIP and a release by the Padres in September and a failure to get signed after that time.

I have done extensive research on the K/BB ratio and certainly, there is some correlation between success in this category and ultimate success in the other numbers, such as ERA.  However, the connection isn't as solid as you might expect and certainly doesn't warrant filtering yourself.

Since 1969, there have been 4,754 pitchers with a K/BB ratio of at least 2 or better, regardless of the number of innings they pitched.  The combined ERA for these pitchers is 3.48.  The combined ERA for everyone else was 4.29, clearly inferior but not enough to cause us to completely exclude this group.  In fact, let me share with you the results of part of that K/BB analysis, which reflects all data through the completion of play in 2002.  Certainly, it tells us something but whether this is enough that we can make exclusions is up to the reader:

K/BB Ratio (all pitchers since 1969) - Combined ERA for All Pitchers

(2 or better) - 3.48
(1.75 - 1.99) - 3.88
(1.50 - 1.74) - 4.09
(1.25 - 1.49) - 4.22
(1.00 - 1.24) - 4.49
(0.75 - 0.99) - 4.86
(< 0.75) - 5.83

As I said, there is an obvious correlation but it is only one key indicator and with many exceptions, we must not confine ourselves to this factor alone.  Think of it is another piece of data that we're considering.

From AskDL June 1, 2003:

Q.  In the previous issue, you covered the K/BB ratio for pitchers.  I'd be very interested in hearing your thoughts on how strong a correlation you've seen, if any, between that ratio and batting performance or in general how the ratio affects your evaluation of hitters, especially somebody like Corey Patterson.  Is his career BB/K ratio the reason you don't see him sustaining a .300+ average.

A.  That's not the case with Patterson.  In his case, it's more that he doesn't make contact often enough, which obviously considers strikeouts but not in relation to walks.  I still like Patterson overall, at least in terms of fantasy value.

Like the K/BB ratio for pitchers, I've never put too much emphasis on BB/K ratio for hitters for a number of reasons.  There are hitters who are great at making contact when they swing but swing at everything.  There are hitters who don't want to walk.  There are great hitters who are fantastic at taking walks but who also strike out a whole lot too (read Jim Thome).

Since I did do the comparison last week for pitchers, I will do the same for hitters here and acknowledge that there is some connection between BB/K ratio for hitters and eventual batting average but I attribute much of that to the K part of the equation, the fact that the more often you put the ball in play, the more likely you are to get a hit.

Regardless, I did have an analysis already done on BB/K ratio for hitters which I'm glad to share with you.  Since 1969, here are the total combined batting averages for hitters in different groups of BB/K ratio (hitters must have at least 300 at bats to be considered in the analysis):

BB/K Ratio and Combined Batting Average

(4 or better) - .316
(3 - 3.99) - .302
(2 - 2.99) - .300
(1 - 1.99) - .283
(0.75 - 0.99) - .275
(0.50 - 0.74) - .270
(0.40 - 0.49) - .268
(0.30 - 0.39) - .264
(0.20 - 0.29) - .264
(0.10 - 0.19) - .256
(0 - 0.09) - .250

There is an obvious correlation but there are more exceptions in each category than can be counted and more than three quarters of hitters fit between the .10 and .99 area, meaning that even with all else being equal, we're talking about the potential difference of less than 20 points in batting average or about 2% of a hitter's batting average.

Where I think the BB/K ratio does come into play, and is worthy of consideration, is in terms of a player's ability to keep his job for the long-term.  In that case, that's where hitters like Patterson have to do a bit more to keep their job.  Since 1969, the average BB/K ratio in baseball for hitters with at least 300 at bats in a season (so we exclude pitchers' hitting) is about 0.67.  Of the 3,308 hitters with at least 300 at bats and a BB/K ratio of at least 0.67, about 74% or approximately three quarters of these players returned to get at least 300 at bats in the subsequent season.  Of the 3,789 hitters with at least 300 at bats and a BB/K ratio of worse than 0.67, about 68% of these hitters returned to get at least 300 at bats in the subsequent season.  As you can see, the difference isn't huge but it is meaningful and it seems to demonstrate that if a hitter has a below average BB/K ratio in a season, it reduces their chances by about 5-6% of returning to a full-time job in the subsequent season.  Of course, many of these hitters may have voluntarily chosen not to return (read retirement).

Overall, and while the correlation between batting average and BB/K ratio is obvious, I am much more interested in K/AB in terms of the impact on both short-term and long-term forecasting.

From AskDL August 3, 2003:

Q.  Is there a closer in Toronto?

A.  Apparently there is not though I still think the Blue Jays are going to eventually give Aquilino Lopez a shot at the role for 2004, if they don't go out and get someone to help them there.  Blue Jays' GM J.P. Ricciardi does not overvalue established closers and so if he did go out and get an arm, I would expect it would be an established reliever who has been stuck in a setup role with another team, as opposed to a 30 save type.  Lopez has been pushed to the majors by virtue of being a Rule 5 pick and he's older than you might expect (28), which is why he was even available in the draft to start with.  I see Lopez as the best reliever currently on the Blue Jays' staff, which isn't saying a heck of a lot.  By default, Juan Acevedo has had the most success there in his recent career and so he's likely to get the most saves the rest of the way.  He is not technically a "closer" by the usual use of the word though.

From AskDL September 1, 2003:

Q.  I'm amazed Juan Pierre is coming out at the top of the list of most valuable 5X5 hitters based on performance-to-date.  I would love to have his 57 steals but he gives up quite a bit to Pujols in the other four categories.  Do you really think Pierre is more valuable than Pujols?

A.  While it looks like Pujols would be an easy choice if you could go back and get either player's season-to-date total on your roster, if you take a typical 5X5 fantasy team and add in Juan Pierre, you will move up slightly more in the standings, on average, than if you add in an Albert Pujols, even though both are great players.  I know it doesn't seem that way but this is an era where the steal is much more scarce and consequently, much more valuable.

To put Pierre's 57 steals in perspective, through this past Thursday, he had stolen 2.7% of all the steals in baseball this year.  To equate that to home runs, it would be the equivalent of a home run hitter having 116 home runs right now (that's 2.7% of the number of home runs hit).  If an imaginary hitter could have 116 home runs and still be weak in the other categories (which obviously could only happen in theory), that would make him an exceptional player who would move your team up in the standings dramatically, even more than a Sheffield or Pujols.

If you disagree, by all means use your own valuation model rather than mine as there are hundreds of different ways of valuing players and many bring a completely different perspective to the table.  The model we use was developed while analyzing what is now thousands of fantasy league rosters and it has been extremely reliable at estimating which fantasy roster should have been the winner when adding up the sum of a fantasy team's parts, compared to how they actually did.  In this era, 57 steals is in astounding number and is comparable to 116 home runs, 481 RBI (again 2.7% of the runs driven in) or 505 runs scored (2.7% of the number of runs scored).  Do that in one category and even if you do nothing else, you're moving ahead of everyone else, even if you do little in the other categories.

By the same token, the scarcity and concentration of totals is the reason that closers are and should often be more highly valued than starting pitchers.  It's not that I like that things are this way but the facts are there that these sorts of players do win leagues and historically, the evidence is there to support that.

Interestingly, Pierre is one of only two top-ten hitters who shows no power in the fantasy value model with Carl Crawford being the other.  The rest all show some pop.

Q.  I am trying to find out what percentage of players in each age group return to the majors the following season.  In other words, how many 40 years olds, for example, come back to play another season at 41 and can you tell me this for each age group?

A.  I actually have done a minor study of this and discovered that there was a significant difference between hitters and pitchers where hitters were more likely to return to play the subsequent season.  Remember that there are many reasons why a player doesn't return to play the following the year.  They might be injured.  They might be in Japan.  They may voluntarily retire.  They could go unwanted on the free agent market.  They may be one of those fringe players who gets called up to help a contending club in September, fully aware that they are unlikely to get another shot again.

Having given you all those examples, let me share with you the results of that study I did on all players from 1969-2001.  The table below tells you what percentage of players in the age group in question came back to make at least one appearance the following season in a year after they made an appearance in season one.  The age group refers to the player's age in the middle of season one.

Take it with a grain of salt for as I've tried to emphasize, there are so many reasons why players don't make it back to the majors in season two.  This does seem to be the answer to the question you asked anyway:

Though I've reminded you of the limitations of asking such a general question, there is a clear trend here for as you can see, the older the player, the more likely they are to disappear from baseball.  Looking at this same data in a graph makes the obvious trend much clearer.  Though the two lines appear to have similar shapes, it is merely coincidence in the later part of the curve, where there is limited data:

... and for those who are interested, our new issues of "AskDL" begin running in the first week of January!