The Best of Ask David: Volume 3
published November 9, 2005
by David Luciani

Editorial Introduction:  I've made a tradition the past three years of starting off our new essay publishing season by compiling a "best of" collection from the Ask David series from the season just completed and every time I do it, I suspect I improve at picking out the right topics.  I don't necessarily like for this to become a place where I can simply edit my best advice and remove my worst (though even the best efforts to avoid that will always be influenced by the natural bias a writer has for his best projections) but rather, I like it to be a collection of the sections from previous Ask David columns that either generated the most positive feedback at the time (i.e. people told me they liked the way I said something or the idea behind it) or that I feel offered a particular perspective I never really had elsewhere in these pages.  As always, I limit myself to the previous year when selecting these and I try as much as possible to preserve the entire comment I made rather than just picking out what sounds good and discarding the rest.  Often, I've inserted a follow-up editorial comment which appears in blue, like this one, so you can know it wasn't in the original issue.  We'll soon be publishing our first issue of the new volume and as always, you can send your questions for consideration to:

Enjoy this annual collection of some of the highlights from 2005's column and I look forward to your questions for 2006!  --- David Luciani, November 9, 2005.

From issue published November 7, 2004:

Q.  What are Gavin Floyd's chances of being a permanent member of the Phillies' rotation in 2005 and what kind of ceiling do you think he is capable of reaching as a major league starter?

A.  A couple of years back at least, I remember that Floyd's absence from my top prospect list generated plenty of negative feedback from readers and I responded to his deliberate omission from the list in detail in an Ask David response, which you can read at our site in the "Best of Ask David: Volume 1" column if you want to see what I said.

While admittedly Floyd has moved up slightly on my list, he still hasn't had the minor league performance to convince me that he's the top prospect that so many think he is.  He walks too many hitters to be successful (even non-translated, he walked 46 batters in 119 Double-A innings in 2004) and while five starts at Triple-A don't give me enough of a sample to be highly confident, his 4.98 ERA over those five starts for Scranton/WB didn't promise much for 2005.

So, in terms of him cracking the rotation in spring training, I'll be surprised if he does, mostly because the Phillies will go into 2005 believing they're a contender and thus are less likely to take risks.  As for his long-term, he's still young enough (he starts 2005 as a twenty-two year old) to develop the skills he'll need to compete at a high level but I take the same position I did a couple of years ago and that is that while I believe he will eventually pitch in the majors, I know that my opinion of him is lower than the average opinion out there.  It's not a criticism of his skills and perhaps it's my own lack of ability to see what those who rank him as a top prospect see.  To me, his stuff looks ordinary and the minor league performance so far hasn't convinced me otherwise, though his performance at Double-A Reading this year (which is a fairly neutral park and if anything can even boost home runs a bit) was a step in the right direction.

I'm sorry but on this one, if you're hoping to hear me say that he's Cy Young, you'll have to ask those who rank him at or near the top of their prospect lists.  You have my opinion anyway.

Floyd didn't contribute much in 2005, going 1-2 with a 10.03 ERA in 26 innings.  My opinion of him remains intact and while I have missed on many prospects over the years, as I'm quick to admit, I'm starting to think that my low opinion of his ceiling several years back may have been justified.

Q.  This past year, I entered September comfortably in first place in my fantasy league but ultimately finished third.  This also happened to me a few years ago when I had an almost insurmountable lead on September 1st but finished second.  I don't manage any differently in September but my team seems to frequently be outperformed.  Any advice on September strategy?

A.  I suspect you answered your own question here and that is that you "don't manage any differently in September."  I believe you have to manage differently in the final month because the majors manage differently.  For starters, a lot of fantasy GMs will only reserve or release players if they get placed on the disabled list and from April to August, that can be a good strategy in that you're sticking with the players you picked and replacing them if they get hurt.  The problem is that in September, they often don't get placed on the disabled list by their big league club (because the club already has an expanded roster) so you might miss that they're hurt and thus, leave an inactive player active on your roster.  This is a mistake that's virtually impossible to make from April to August but easy to do in September.  If you want to find out if you made the mistake, look back at your September performances to see if you got little or no contribution from an injured player who remained active in the majors.

Another problem with September is that many real clubs are facing inferior competition as they go up against September call-ups who aren't ready for the majors.  Your opponents are exploiting this and you need to be too.  Particularly in the case of starting pitchers, you have to look for those starters who are going up against non-contending teams down the stretch because, on average, you can see an ordinary starter look like an ace if he got to face the September call-ups of Toronto, Detroit or Arizona this past year.  This is just as true of closers too, who can accumulate more than their fair share of saves simply by facing teams who don't expect to win much in September.

The flip-side of the non-contending team is the one who has clinched a playoff spot.  If you had plenty of Yankees or Cardinals in the final week or two of the season, you might discover that these teams were giving a day off to players who would ordinarily be in the lineup everyday.  Particularly if you have even weekly transactions, you have to watch for this because you can't blow away your opponents if you're getting 4/7ths of a week of performance from your players.  You want to have players active who both have a need to play and who are up against competition against whom they can perform well.  It seems like common sense but in September, the schedule is everything.

One last thing to watch out for is how your opponents are managing their rosters in September.  By the end of the season, roles are obvious and clear and you can get away with speculating throughout the first four or five months of the season but by September, everyone in the league knows exactly who has a job and who doesn't and so if you're counting on a player to break through in a given week, and this may work throughout most of the season, chances are good you won't see it happen in September if it hasn't already.

Finally, the season is a long run and it's possible that you just didn't have the best team.  I always hate to say that but it can be that you just had the "rolls of the dice" going in your favor for much of the season and it balanced out.  One year, I went into a season believing I was rebuilding (in a 5x5 AL-only league) and was pleasantly surprised to be in first place by a wide margin, even as late as August 15th or 20th.  By the time the season finished, I finished in third place.  Initially, I looked back to see what mistakes I had made in September only to recall that I went into the season with low expectations.  The more I looked at my roster and the other two teams that finished ahead of me, the more I realized that I had overachieved for much of the season and the luck finally balanced out.  I took my third place finish with pride that I had done so well in a season when I didn't expect to contend.  This may not be the case with you but the story will be relevant to someone out there in a similar situation as you mentioned.

From issue published November 28, 2004:

Q. I recall that you were low on Danny Kolb last year and he had an outstanding season. What went wrong and what do you think his prospects are for 2005?

A. There are times when a forecast misses and I can look back and say I missed for this or that reason and move forward with new information to improve the process. In the case of Kolb, I can't do that. In fact, the elements of the forecast that led me to believe he would be ineffective were all there, particularly his projected lack of strikeout skill. He struck out 21 batters in 57.1 innings and yet managed to allow only 50 hits and 3 home runs and a 2.98 ERA. Like I discussed in a recent column about so-called DIPS theory (Defense Independent Pitching Statistics), while I don't entirely subscribe to the idea that pitchers can't control what happens after contact is made, I do believe there's merit to the concept because consistently, pitchers who prevent batters from making contact do well, historically, and pitchers who don't, at least on average, suffer with the rarest of exceptions.

If we look at Kolb's season in some detail, his rate of non-home run hits per ball in play given up was 23.9%, which is an excellent and even occasionally top ten type performance. If we look back at his rates over the previous three seasons, he allowed rates of 27.6%, 25.5% and 28.9%, which are more in line with my expectations, given his stuff. In fact, I believe he's closer to a 28% or 29% on this scale and that's why I typically project him to allow more than a hit per inning.

In his favor, he does keep the ball down well and thus rarely gives up a home run. Regardless, with the hits per ball in play being out of line with what I believe his real ability is combined with what I believe has been a slight overachievement the past couple of years in his ability to prevent home runs, I am prepared to project a decline in performance in 2005.

I show this comment despite my being wrong about Kolb in 2005 (I later projected him to have a good season saves-wise and a much better season performance-wise than he did) because it may explain what happened to him as I had projected an ERA above 5.00 in 2004 but not in 2005.  The comment about DIPS theory was one that generated much feedback and so it's why the excerpt makes it into this collection.

From issue published December 12, 2004:

Q.  I'm really disappointed at your Jeremy Bonderman projection.  I have him projected for a top season in 2005.

A.  As the projection notes say, you're entitled to differ with me and in fact, I'd be disappointed if you couldn't find a forecast or two, at least, where you don't agree with what I said.  I was surprised to see that at least four people asked variations on this question, which made it clear that there are some positive opinions about Bonderman's stuff out there so readers may want to consider that there are quite conflicting opinions from what I'm forecasting.  I actually didn't think that Bonderman's projection was out on a limb as much as some others as I'm projecting an 8-10 performance and a 5.24 ERA over 146 innings (he has a career 5.20 ERA over two seasons).  Yes, his 2004 season was an improvement on 2003 and he looked sharp in the final month or two of the season but he's the type of pitcher I believe who can fool a team the first time he faces them and then becomes hittable once they get familiar with his stuff.  In fact, for teams against which he started even just twice in 2004, he had a sub-4.00 ERA for the season against only Oakland, Seattle and Tampa Bay, all three of which were in the lower half of run-scoring teams in the American League.  In his fantastic month of September, he made only one start the whole month against even a remotely contending team, that one against the White Sox on September 19th who were already essentially out of it and would be mathematically eliminated the next day.  In the rest of the month, he pitched (sometimes twice) against only Tampa Bay, Kansas City, Cleveland and Baltimore.

... If you think he's going to have a top season in 2005, by all means please disregard my projection.  My belief is that the turnaround of the second half won't be sustained and as late as mid-August, he was still having a worse season in 2004 than he was in 2003.

This was actually the first question I answered after our first 2005 projections were published and while I underestimated Bonderman and again, I include this despite my "miss" on the forecast, I don't recall a single pitcher that didn't have a true track record for greatness generating so much email of high expectation, not just throughout the pre-season but early in the season when he looked sharp.  Ultimately, the "breakout" year that so many seemed to be projecting never really happened, though his wins didn't hurt.  Bonderman faded and while finishing with a 14-13 record, it went with a 4.57 ERA in 189 innings.

Q.  I can't understand how Mark Prior is forecasted for only 12 wins in 28 starts but Josh Beckett is forecasted for 14 wins in 26 starts.  Prior pitches for a better team.

A.  This may come as a surprise but I'm actually projecting the Cubs to be a disappointment in 2005 and in fact, the implications of the forecast are that the Cubs are projected to be a sub-.500 team, though this does not yet account for the Garciaparra signing and other factors.  As it stands with the first forecast set, I'm projecting the Marlins to be the best team in the NL East.  That explains the wins and this represents a turnabout for me, of sorts, on Josh Beckett, for whom I projected only 10 wins in 2004.  I also believe the Marlins have a superior bullpen than the Cubs, at least as of the date the first forecasts went online.

This was an important reminder I have to make at least annually that the projected quality of a team plays a big part in the forecasted win totals for a pitcher.  Ultimately, Prior won 11 games in 27 starts and Beckett won 15 games in 29 starts.

Q.  Why are you so high on A-Rod for 2005?  It's obvious his skills have declined.

A.  A-Rod was asked about by at least a few readers and admittedly, my forecast may look high but in fact, I'm basically projecting him to perform at the level he did from 2001-2003 with Texas.  While Texas was a hitter's park, it wasn't the only reason A-Rod was performing so well and in fact, over those three seasons, he hit only 9 more home runs on the road than at home, total, or about 3 per season.  As for his so-called decline, he doesn't turn thirty years old until late July of 2005 and I have trouble believing that he's already on the way down.  I think he suffered through an anomaly season in his first year in pinstripes and he will rebound to customary levels in 2005, with the exception of his speed which was actually better in 2004 than we've seen since 1998 and which I don't think will be as good in 2005.  This may surprise you but I've often explained that my forecasts are a mix of statistical observation and scouting and what I observe is that I think A-Rod is actually improving as a hitter.  The new confidence model will reflect the uncertainty associated with this forecast though his low end of expectations is still better than many players' upper ends.  Like my response to Bonderman, all of the forecasts come down to opinions about the players and you have mine for A-Rod and much of his forecast is based on scouting, as much as statistics.

I don't think I need to say much about how A-Rod did in 2005 and the important point here is that scouting plays a big role in the forecasts, even if I apply my scouting knowledge in a statistical fashion.

From issue published January 2, 2005:

Q.  I love fantasy baseball but I'm wondering if there's anything I can do to make it more realistic, other than changing the categories we use.

A.  There are so many things that can make fantasy baseball more realistic but "more realistic" does not always mean "more fun."  Here are just a few of the suggestions over the years that have come up, some of which I've tried and some of which I haven't.  I think a few of these were original ideas but I'm sure just about everything's been tried around fantasy baseball circles.

Try putting in long-term contracts for players rather than simple, escalating salaries.  This certainly makes things more realistic because rather than being asked to predict just how a player is going to do in the upcoming year, in order to get the player, you have to predict how good he will he be many years in advance.  I've tried this in two different leagues and the best version of the rule went like this:  During the free agent auction, you bid just as you normally would except that free agents are broken into groups by which round they go in.  The very top free agents (i.e. the first round) get five year contracts, round two and three players get four year contracts, rounds four through seven get three year contracts, rounds eight through fifteen get two year contracts and everyone below that gets a one year contract.  Since inflation isn't a factor, the contracts have the same salary per season.  This simulates the idea that if you're going to go after a top free agent, he's going to want (or we're simulating his wants of) a long deal and you could end up gambling and eating his contract if he doesn't perform.  Rather than deciding whether to keep a player in a future year, if he's under contract, you have to keep him, even if you're paying him to sit on your reserve list.

If your league doesn't already use waivers, you could implement that so that the player who's signed to a four year unwanted contract could be placed on waivers to see if another team claims him at his salary.  Of course, if no one does, then you're stuck with him still.

If you go this route with contracts, you can also try putting in a rule in your league that when you sign a star free agent, he gets a "no trade clause" that prohibits you from dealing him to any other team except the first or second place team for the duration of his contract.  This is simulating the idea that most real players with a no trade clause are willing to go a team that has a chance to win but otherwise want to stick with the team they chose through free agency.  A group of us here did try this in one league back for a few years in the early 1990s and what we said was that any player who was bought in the first round of the free agent auction in any year counted for this rule for the duration of that player being retained.  It actually caused several problems for teams who ended up out of the race a few years down the line and were still carrying that expensive player whose salary they wanted to shed, sort of the fantasy baseball equivalent of the Bobby Bonilla syndrome.  No one would take the expensive player off their hands whose career didn't go the way the fantasy GM thought it would.  If I recall correctly, someone ended up with Albert Belle in this league at $35 or $40 per season for two years after he was already done.

A friend of mine and I tried another idea in a tiny six person league that was great fun and we meant to return to it but didn't.  Just like real baseball, we linked performance with income so that if a team does poorly every year, its future year salary actually declines.  Basically, what we did was we said that any team that finishes in the top half of the league gets a 10% increase on its maximum salary compared to a year earlier and any team that finishes in the bottom half gets a 10% reduction compared to a year ago.  We even put in an additional 10% bonus for the first place team and an extra 10% penalty for the last place team and we placed no limits on subsequent seasons so that a first place team could theoretically grow right out of the rest of the market.  What we found was that everyone was desperately trying to avoid last place and that teams were fighting to stay in it as long as possible because they wanted to do better to either get that bonus or save the last place penalty.  Since it was a small league and we only tried it over a four year span, we really didn't get a sense of whether it benefited leagues to use and if anything, we were trying to simulate the lack of balance in real baseball.  Most fantasy leaguers try to do just the opposite so that everyone, with all else being equal, has an equal chance to win in any given year no matter how they did a year earlier.

One last thing you can do to make your league more realistic, and this one doesn't require changing your rules at all, is consider using a draft board at the auction/draft.  There's a company (no, I don't have any ownership or make any commission whatsoever off this product) called FJ Fantasy Sports who makes draft boards for fantasy baseball auctions.  It certainly makes the draft more realistic because you can see the players on a clean looking board, sort of like watching the NBA draft.  It makes your auction into more of an event than it already is.  Before I wrote this note, I decided to call Jeff Peters, the owner of the company, to tell him I would be mentioning his product here and asked him on behalf of our readers if he would give a 10% discount to Baseball Notebook readers who buy his product.  He said he'd be glad to as long as you phone rather than order online.  If you want to see the draft board, you can see it at FJFantasy.com but if you want to get that discount for Baseball Notebook readers, phone Jeff instead at 1-866-200-7493 or 1-814-449-1269 to get yourself set-up.  I don't pretend to know his regular prices but these boards are high quality and they'll make your auction/draft seem like something you would watch on ESPN, not to mention that they make it easy for everyone in the room to see who's been drafted and where.  As I say, I'm not plugging them for any other reason than this will really make your draft seem like a true sporting event.

Several readers wrote me to say that they had tried at least one of my suggestions I gave here for the first time in 2005.  I'd be curious to hear whether it improved their fantasy baseball experience.

From issue published January 16, 2005:

Q. What do you think of the Beltran signing in New York and where do you think Delgado will end up?

A. I haven't yet finalized a Beltran forecast for him as a Met though I will be doing so by next week's projection update, obviously. I'm surprised that Beltran was considered the best free agent on the market. He's a good player and had an outstanding year but I have to wonder whether he really helps a team win more than a Carlos Delgado, who is going to sign somewhere for significantly less money and fewer years. Before last year, Beltran had never hit 30 home runs in a season, still has only two .300 seasons in his career and has a career on base percentage of .353 and a slugging percentage of .490. He is one of those few base-stealers who really does contribute quite a bit by virtue of his stealing alone because he rarely gets caught but when I compare those numbers to Delgado's, I remain surprised at how the market apparently perceived Beltran. Delgado's got a career on base percentage of .392 and a slugging percentage of .556 and while he is about five years older, he's not looking for an eight or ten year contract. I just believe that he contributes more to winning than Beltran does and thus, should have been perceived that way. As good a player as Beltran is, I believe his late-season performance in 2004 created an entirely false impression of his overall skills which are still better than he was in his early Kansas City years but not as good as those we saw in September. Of course, fantasy-wise, he no doubt has big value because of his running game.

As far as where Delgado will sign, I'm not sure how much money the Mets have left in their budget or whether they're about to move someone to free up more. I think Baltimore's out of the race here and that leaves Texas, the Mets and Florida. Between those three, we can't know what conversations are going on behind closed doors. All three could legitimately be called a "contender" in the sense that will satisfy Delgado's desire to play for one and money-wise, it would seem Florida has the most free in their budget. So, the only guess I'll make is that Baltimore's out of the running on this one as they don't seem to be making the same effort as these others.

I actually wrote a lengthier discussion about Beltran in December 2004 that proved to be one of our most popular essays of the season and it briefly alluded to my own salary model for building a successful "real" major league team.  That article was called "Why Carlos Beltran Could Be Worth $20 Million a Season" and is likely headed for our archives section of most popular essays.

Q. I was wondering if there are any articles or your thoughts on the idea of position scarcity and do you use it or discount it in terms of auction format? For example, in my league, catchers routinely go for almost double the projected dollar amounts from your Fantasy Domination Ranking Sheets.

A. I actually used to believe in it and was even a proponent of it until around the mid to late 90's when we did a study on several hundred (and subsequently within a few years, several thousand) real fantasy leagues worth of data. What we found was that with perfect hindsight, position scarcity simply did not work and it wasn't even close. I know there are readers who disagree with me on this and I've heard the many good arguments on the other side but a home run is a home run whether your catcher hits it or your first baseman hits it. One argument goes that you should pay more for the catcher because he gives you more production out of a spot from which you normally wouldn't get much production. In real baseball, this is absolutely true and I'm completely a believer in real baseball position scarcity value but it's because the player makes a defensive contribution that doesn't exist in fantasy baseball.

When we did this analysis on the many fantasy leagues, we tried drafting based on inflating prices for players at weak positions and we did it with perfect hindsight, knowing exactly how players would perform. We even knew, with perfect hindsight, exactly how weak each of the positions really were. We had the computer simulate billions of drafts using these many leagues of data and with a third of the teams doing absolutely perfect position scarcity analysis and valuation and the rest of the teams ignoring it. Even with perfect hindsight, the position scarcity teams dramatically suffered in the standings, often ending up with middle-of-the-road teams that got reliable production out of the weak-hitting positions like catcher and second base, but suffering overall because it either had to try to find bargains at the good-hitting positions or because it ended up paying more for the production of the 20 home run catcher as an opposing team paid for the 20 home run outfielder. If the positions are weak, then it's okay to fill them, on average, with weak hitters. If you get lucky and can buy Alfonso Soriano at the same price you would have paid for him to fill your DH spot, great. If not, you move on and accept it.

After that study, and the sequel to it, which had even more data to work with, it seems hard to believe that I used to believe in this idea in fantasy baseball and I know there are many people who still do. It doesn't work for me and I've tried all of the valuation adjustment methods out there, both using hindsight analysis and for forecasting. The most common argument I get is that the guy who does pay more for the 20 home run catcher says that "it's easy to fill a roster with good hitting outfielders late in the draft" and I argue otherwise. They're only good in terms of a distorted perception based on the real baseball population rather than your actual fantasy population.

I can tell you that this is an area I know I differ on compared to some other fantasy experts and perhaps they're right and I'm wrong. My analysis tells me it's a good way to finish around the middle of the pack and if I'm wrong about that, then somehow I've found fantasy success most of the time by doing otherwise. If you believe that a home run from your catcher is worth more than a home run from your first baseman, then you are very much on the other side of this debate from me. At least you know where I stand.

I remain aware that my opinions on position scarcity are not widely accepted by many well-respected fantasy players.  I present my views for your enjoyment and ultimately, the decision rests with the fantasy GM on whether they'll subscribe to the concept.

Q. Can you please discuss why you think Melvin Mora's production will drop so much?

A. One of the challenges of forecasting is knowing how much weight to place on a player's most recent performance compared to what he has done earlier in his career and Mora certainly is a strong example of this challenge. Sometimes, my decision about how much weight to place on a player's latest season or two will be based on my scouting of his skills and other times it will be a purely statistical analysis. Mora's case is a combination of the two, more statistical in the doubles and home runs category and more of a scouting observation in the singles column.

I often avoid answering these questions because some readers tell me my answers are unnecessarily overly detailed or statistical and I don't mean for that to happen but it's the truth behind the forecast and the only way to give you my thinking is to get into the details. I hate to do that but in some cases it's necessary because those detailed numbers, the ones that understandably turn off readers who prefer more scouting type "meat" in an answer, are the sort of things I'm actually considering when making a forecast. When I give you a forecast, I'm trying to save you all this sort of thinking but because Mora's been asked about so often, I'm going to give you some of the details, albeit summarized though still more lengthy than I typically like to give in this column, in the hopes that the readers who are statistically inclined will appreciate the response and those who aren't will skip over my response here and move on to the next question.

My latest forecast for Mora is a .277 hitter with 17 home runs, 63 RBI and 490 at bats over 125 games, which no doubt is a significant drop-off from his 2004 performance.

The largest contributor to the drop is the projected playing time at only 125 games. Mora's strong 2004 season came in just 140 games and he's never played 150 games in a season in his career, much because he didn't break through will a full-time role until he was almost thirty years old but just as much because he has suffered an injury every year since he has. Last season, he landed on the DL because of constant hamstring problems and a sprained foot. In 2003, he missed time because of a hand injury, a face injury, hamstring problems again and he missed most of September with a torn ligament in his knee. I consider him a high injury risk and he's a bit older than some think, turning thirty-three next month. So, I'm forecasting him to play only 125 games instead of 140 and that's a big factor in the drop in production.

The RBIs and other peripheral numbers like runs scored are directly, and I dare say, obviously tied to the primary portion of the forecast which is his projected batting average and more specifically, his projected number of singles, doubles and home runs, with projected triples being a non-factor in his case.

Now, the projected drop in batting average is related to his contact rates and success at getting hits per time he puts the ball in play. I'm projecting him to make contact in about 81% of his at bats. Last year, he made contact in about 83% of his at bats and the year before about 79% and for his career, he's made contact about 81% of the time.

I don't believe he can maintain either the high doubles or home run rate he showed last year, which was a double in about 9% of the balls he put in play and a home run about 6% of the time he put a ball in play. Both are significantly higher than his 2003 season and at thirty-one and thirty-two years of age those years, he shouldn't be improving that much.

In the history of baseball, there have been 94 players (minimum 300 at bats in both seasons) who at thirty-two years of age, based on the age they were for the majority of the season, saw a 50% increase in doubles per ball in play compared to when they were thirty-one, as Mora did from 2003 to 2004. Of those 94 players, an amazing 80 of them or 85% saw an immediate reversion to lower levels in the third season, the season in this case that we're trying to project. The average doubles ability in year three for all players in the study fell off to exactly 80% of the doubles ability the player showed in year two, even including those players who didn't revert. My current projection for Mora's doubles rate is that he'll drop to about a 7.3% doubles hitter per ball in play or about 81% of the ability he showed last year.

The reader might wonder why so many of the players in the study I'm demonstrating here fell back to Earth. The reason is, as many would expect, that players in their early thirties typically don't improve at hitting doubles. They can show flashes of brilliance for a year or two but by the time they hit their early thirties, they're already combating the inevitable effects of age. Granted, some do improve and that's reflected in this no doubt oversimplified example by the 15% who didn't revert at least somewhat to lower levels. In fact, my projection for Mora's doubles ability is an improvement over the former level he showed heading into 2004, which had a career doubles rate per ball in play of just 5.16%. In other words, I'm accepting much but not all of the improvement he showed in 2004 because I do believe that despite his age, he improved some. As I said before, the challenge of a forecaster is knowing how much to accept of a single season performance and in Mora's doubles ability, I've accepted quite a bit of it in my forecasted doubles per ball in play percentage of 7.3%.

In the home run column, we see a similar explosion on his part to a career-high 27 home runs. On a per ball in play basis, Mora homered about 6% of the time he put a ball in play or about a 20% increase over 2003 levels. Again, looking at players who were thirty-two years old for the majority of the season, there have been 361 hitters in the history of baseball who were thirty-two years old in year two and who improved by 20% or more in their home run per ball in play rate from year one to year two. Of these 361, an amazing 248 of more than two thirds (about 69% to be more exact) did not maintain their newfound year two home run levels in year three and the average result including those who maintained or exceeded their year two performance was a drop to about 87% of the ability shown in year two. In Mora's case, I'm forecasting a drop in home run rate per ball in play to about 4.2%. His career rate is a lower 4.0% and I'm accepting some, but not all, of the power he's shown the past couple of years.

As he rarely triples I don't need to get into that but that leaves the single. His singles per ball in play rate in 2004 was 26% and his career rate is just 23%. Of almost any category I examine when building forecasts, the singles per ball in play rate is one of the most consistent across a player's career, even as his skills improve or decline and also, it's one of the ones where scouting takes more of an effect than statistics. I used that rate to explain why I was projecting Ichiro to come out of his season-long slump last May (and explained it in this space at the time) and why at the end of April 2004, I didn't think Ronnie Belliard would maintain his newfound batting average skills. The wonderful thing about this category is that it gets to be such a large sample because there are many balls in play off the bats of players. It's one of those few categories where we get to see a so-called "long run" over the span of just a few seasons and it's also one of those skills that I think can be scouted based on the way a hitter approaches his plate appearances. I don't know if I'm at least slightly bridging the gap between scouting and statistics here but I do think that's been the case and ironically, it's a statistical analysis that shows that my scouting methods in this area work as well as other ones I could have used based on raw numbers.

I'm projecting Mora's rate to drop to about 22.3%, a hair lower than his career rate and this is based on consideration of my age effects model in combination with his scouted performance, including consideration of last year. There were only 23 players in baseball last year who had singles per ball in play rates of 26% or better (and Mora was one of them) and when you have such a small group attaining this level, I have to consider which players will maintain such a high level and which ones are having a few balls roll their way. Mora is a player who had a singles per ball in play rate of just 17% as recently as 2002 and while he jumped up to an amazing (but unsustainable) rate of 28% in 2003 and then fell back a bit last year to 26%, this is the area where I don't believe he has new skills as the career outweighs the past two seasons for me. As I said, this rate is one of the few where career averages often are more reliable than the past season or two and this is the area where scouting absolutely does enter the equation more than statistics. My read of Mora's ability is that the way he hits the ball, which is still efficient, is not a style that will enable him to maintain anywhere close to a 26% singles rate per ball in play over the long run. It's the same sort of scouting that led me to predict similar drops in the singles per ball in play rates for Luis Matos, Marlon Byrd and Scott Podsednik in 2004, all of whom fell from levels above 26% in 2003 down to levels below 23% last season. Almost every such player who had such a rate in 2004, other than Mora, who posts such a high rate are singles hitters who rarely swing for the fences and in fact, Mora had the most home runs of any of the 23 players with a 26% singles rate in 2004. I'm not penalizing him for his production. I'm simply saying that a hitter with his type of aggressive power swing can't maintain that singles rate and if I'm wrong about this, it will account for about 30 points of additional batting average, turning him into the .277 hitter I've projected into something closer to a .310 hitter.

Even if I'm wrong about the singles ability, though, the doubles and home run rates that I'm even more confident about will reduce him to being no more than about a .310 hitter and certainly not the .340 hitter we saw last year. The only way he hits .340 this season is if he's among the 15% of players who really do improve their home run rates at thirty-two years of age. I say he isn't and if you think he's among those 15%, this is one where you should substitute your own forecast in place of mine.

I don't recall ever getting into such detail about the "why" behind a forecast and I wish that I had the time to write an essay or response of this length for every player but it simply isn't possible.  It can't be farmed out to other writers as I'm making the forecast but what you saw here was a documentation of the method behind my Melvin Mora forecast.  At the time of the response, in January, the forecast listed here was for Mora to hit ".277 with 17 HR, 63 RBI and 490 AB over 125 games" as the response says.  As it turns out, Mora stayed healthy much more than I expected, hitting .283 with 27 home runs, 88 RBI and 590 AB over 149 games.  I was wrong about how healthy he would be but in terms of the actual ability, he was right in line with expectations.  Since the response focused more on ability than playing time, I thought it was fun to look back at it now.

From issue published January 30, 2005:

Q. I am confused about how you view draft inflation.  If my goal is to pay 78% of the actual value of my team, when draft inflation kicks in, doesn't that mean I should be paying 78% of that inflated value?

A. This question was passed on to me by our discussion forum administrator because apparently there was some debate or curiosity in the forums what my position is on the issue was.

You should absolutely not pay 78% of the inflated price and in fact, in the essay I published called On Paying 80% of Projected Value which in some ways is linked with the concepts discussed in Perpetual Leagues: Making Your Keeper List, I argue that you control your inflation regardless of what's going on in the draft.  The idea is that if you can get bargains early in a draft or better yet, with your strong keeper list, you can afford to overpay for players during the draft.  In fact, I frequently run into this into the mid or late parts of a draft where my patience in the early stages leads to bargains about half-way through the draft.  Using the calculation methods I describe in that first column, I run into a position of being able to legitimately execute my strategy by overpaying for players late in the draft.

You know, a lot has been written about draft inflation by many writers and I know I'm in the minority but I don't believe in overpaying for players just because there's too much money left to be spent by everyone and too little talent.  No doubt draft inflation exists but just because it exists doesn't mean you should throw your money around the way everyone else is by overpaying for players.  Equating this to real life, I know it happens in baseball where thin free agent markets cause teams to bid for players as though they were twice as good as they are but the problem with this is that you're not achieving your precise goals.  Your goal is to achieve a roster that costs you only 78% of its non-inflated projected value.  In a $260 salary league, that means you need about $333 in projected value to have a strong chance to win that league.  If draft inflation is forcing prices up during the draft and you pay related to that, then once the draft is over, your roster has no memory of the draft itself and that player who was really projected to be worth $30 but had a so-called "inflated value" of $40 is again, still worth only $30 to your team.  If you paid 78% of $40 (i.e. about $31) then you've overpaid for the player.

If you've followed the methods described in that essay about the 80% rule (which has in fact more accurately come to be called my "78% rule") then your keeper list will dictate your own team inflation and I usually run into a situation late in the draft where I can afford to overbid because I'm ahead of my pace to achieve the 78% target.  Exceptional keeper lists, in fact, can have you go into the draft able to overpay from square one.  One year, back in the days of playing on CompuServe's old forums, I went into a season with such a good keeper list, a lot of it the result of tremendous luck because of closers who emerged from my middle relief corps the season before, that I started the draft able to be bidding more than 30% above the projected value of the player.  I was able to snap up several superstars from square one because of this keeper list, using the exact principles I talk about in that essay about the 80% value goal.

By the way, if you're in a league that doesn't have keeper lists, then you go into the draft after the 78% goal from square one.  The key to success here is getting at least a few bargains by the middle stages of the draft so that you can then afford to overpay for the last players on your roster.  This takes not only good attention on your part but also a willingness and discipline not to overpay for a player you really want if he doesn't fit into the model.

The most common mistake I see in both keeper and non-keeper leagues is that either people break their discipline because they feel they aren't getting any good players or worse, finish the draft with money left over because when it came their turn to call out a player, they selected players who had too wide an appeal and thus, had no chance to actually end up on their roster.  If Albert Pujols is the player everyone in your league wants, then when it's your turn to call out a name, while it is advantageous to call out his name if you want everyone to start overpaying for players, in order to secure the bargains you need, you need to call out names early where you think the highest bid anyone will make is way lower than your 78% target for the player.  A great example of this is Johan Santana last year, who showed up as being projected to be worth $32 in a medium-sized 5x5 fantasy league on our fantasy domination ranking sheets last year.  If you're the guy who would have called out Santana's name in an auction, the nice thing is that you'll get the last bid and can top things by just $1.  That means that if it went around the room and the highest bid had hit, say, $21 or $22, you could have bumped things up by $1 and gotten a bargain that then inflates the amount you pay for other players.  You must take advantage of your opportunity to name who's on the block when it comes up and you must be naming players you believe you have a chance to obtain within your current bid ratio (as described in those essays I mentioned above).  If you can do this consistently, you'll pick up bargains frequently and throughout all stages of the draft.

If you're in doubt about what the general perception is about player values, I strongly recommend the site mockdraftcentral.com.  They're not affiliated with Baseball Notebook in any way but they'll tell you exactly what the general population thinks about players right now, even down to average dollar values being paid for each player in mock drafts during the past week.

Q.  I noticed on your January 23rd projections that Johnny Estrada was only projected to score 41 runs and have 48 RBI.  Is this related to the right and left field holes the Braves currently have?  He also has fewer projected at bats than last year.  Do you feel that he is going to platoon with Eddie Perez?  I am trying to figure out his value in the new look Braves.

A.  Actually, the January 23rd projections were already able to account for the signings of Raul Mondesi and Brian Jordan and so Estrada's run and RBI production accounts for the new-look lineup, though I'm not projecting Jordan to stay healthy for the full season.  I do believe Estrada played a bit more often than we should normally project for a catcher, especially one who never played even close to that amount before, and the projection you quote has him playing 128 games and getting 342 at bats, which looks reasonable to me given his limited history.  Yes, Eddie Perez does get more of his share of the playing time here with 91 games and 212 at bats projected (compared to 74 games and 170 at bats in 2004).

The drop in playing time for Estrada is also tied somewhat with the projected drop in performance, too, as I'm projecting "only" a .284 hitter compared to a .314 performance in 2004 and particularly in the doubles column, I'm expecting a drop, which corresponds to the projected drop-off in RBI.  On a per at bat basis, his run scoring remains about where it was last year.

Among hitters with at least 450 at bats last year, Estrada's RBI production per total bases was 8th highest in the National League and hitters tend not to repeat on this list year after year.  I suspect it's because RBI and total base totals are so closely related to each other.  In fact, it's fun to look back at others who were among the top ten on the RBI/TB list in 2003 because virtually all of the top ten saw tremendous drop-offs in RBI production in 2004, not necessarily because their power or extra base numbers or even their overall raw performance dropped off but because it's hard to maintain high rates of RBIs per total bases because those who lead the league are usually there by a combination of their skill and a bit of good fortune.  Estrada, as far as I'm concerned, falls into that category and I'll be quite surprised if he drives in 75 runs this season.

The important point here was for the reader to understand the link of projected performance to projected playing time and also how R and RBI from one season are not necessarily the best future predictors of those categories for future seasons.  Though it wouldn't have changed the point had he performed much different, for what it's worth, Estrada hit .261 with 31 runs scored and 39 RBI in 357 at bats over 105 games.  I was wrong about the doubles drop.

From issue published February 13, 2005:

Q.  I just found your site the other day and in your archives, there is an article in which you suggest your projections are 95% accurate within a "reasonable range" of accuracy. I'm wondering what you mean by that. It appears common ground among all other forecasters is that 70% is about the best that can be done - and that is using a relatively broad interpretation of accuracy (see the widely published essay Ron Shandler wrote this year regarding projective accuracy, for example). How is it that you can claim to be so much better than everyone else? Or, why do you use such a broad definition of accuracy to the point that the term "accuracy" appears to lose its conventional meaning? Just wondering.

A. This is an excellent question and there are so many aspects to it that I would like to address here.

First, your note motivated me to remove that old article from 2002 because when you wrote me, I went back to re-read the essay you were talking about (which was about a different topic and not forecasting but mentioned the 95% reasonable range) and I see that I didn't there offer a more detailed explanation of the so-called "reasonable range" of accuracy that we defined way back in 1998 when our site was new. I've since grown to strongly dislike talk of accuracy rates because, even as that removed article pointed out, not a single projection will technically be accurate. I can be precise here and tell you that 95% of forecasts will be within two standard deviations of the forecasts and I believe we're the only site that publishes our annual standard deviations in every category when the season ends.  There's no secret about what the accuracy rate means.

By no means do I claim (nor should anyone claim) that projections are 95% accurate and if you got that impression from that essay, it wasn't my implication.  True accuracy means a perfect projection and there won't be a single player who has a forecast that precisely matches the eventual outcome, barring extreme coincidence and a tremendous amount of luck.  It is true that we came within a triple one year for Brady Anderson but that was when we only published eight categories for hitters rather than the 20+ we project now and as I've said, that was a lucky outcome.

In fact, the reasonable range definition came about when we moved from print to the Internet and in 1998, I opened up an early discussion with readers about defining precise ranges that would constitute a reasonable range of accuracy. Ultimately, after engaging our new Internet readers in discussion, and many of them had carried over from our print media days, I ultimately published my conclusion (which I have since distanced myself from at least two or three years ago) that the definition I was after for a reasonable range of accuracy is a projection that does better in every category than a forecast created using any combination of previous year's data available through weighted or multi-year averages or single seasons. I conceded even then that the reasonable range wasn't perfect but to me, it constituted a real value for forecasts because that meant that you were producing a forecast that was more useful than any number you could get from just looking at surface history.  Though our annual standard deviations do now meet this goal every year in every category (i.e. our standard deviation in each category is always more accurate than any three or two or single year average you could create from published data), I abandoned this definition in late 2002 and I hadn't recalled, until you pointed it out, that there was still an old essay in the archives - about a different topic however - that alluded to the 95% confidence level and the "reasonable range" which I did not want to use anymore because I don't like it. You can see precisely how accurate we are each year by looking at our annual published standard deviations but I don't like references to percentages because they really are meaningless because everyone's tolerance level for accuracy is different.

You will still see me use the phrase "a 95% range" in other contexts, though, as it is a frequently-used confidence level when I decide whether to exclude the relevance of data from a player's past and this is a common cutoff level in statistical circles, particularly in election polling.

As far as Ron Shandler's recent article about forecasting goes, I did read it and thought it was excellent and I told him so a few weeks ago.  I like to think he wasn't talking about any flaws in what we do here because we have never in an advertisement claimed to be "95% accurate" for example, as other sites do. I disagree with Ron on some minor points but as I say, I don't think he was criticizing any of the claims we make here as we present data and let the reader decide precisely how tolerant they want to be by looking at our standard deviation tables. By all means, of course, you could write Ron and ask him as I'm confident that he and I take a quite similar methodic approach to forecasting and while we often yield different forecasts from our methods, I highly respect what he does and have never once claimed to be more accurate than him (or any site for that matter). As Ron so correctly points out, people who do comparisons of forecasting systems almost invariably conclude that theirs is the best through whichever method they pick. I rather prefer to simply say, here are our standard deviations - you decide if they're valuable to you and you go ahead and compare them to any other method out there. There are so many reasons that I don't compare to other published sets, first and foremost of which is that there isn't an agreed-upon method to compare but just as importantly, we don't all publish the same players or at the same times and it would be unfair of me to compare my forecast sets to any other without a common publishing date and an agreed-upon pre-established criteria for rating the forecasts that all other prognosticators united behind.

In fact, Ron is too polite to say it in his essay but the rating method he describes is one that was exactly used in one of the Baseball Prospectus books to demonstrate that their so-called Pecota system worked better than other methods they polled, including Shandler's forecasts. The flaws in the comparison were admitted right in their book but to me, those flaws were so apparent that it made the comparison pointless. For example, they only compared players that all sets published. As Ron points out, that isn't fair. In fact, that punishes a set like ours which publishes projected September call-ups because other systems aren't being called upon to project everyone. Ron also correctly points out (again without naming BP) that they used an OBP+SLG comparison method to rate accuracy which, again, is not necessarily the best choice. As I say, I can't say for sure that Ron was talking about the Baseball Prospectus comparisons but the properties he describe were exactly in that comparison published in the introduction pages in their book. I highly respect and enjoy the work that the BP folks do.  That writers of theirs are now working in senior baseball operation positions speaks volumes to the great work they do.  But even with them conceding that they weren't making a claim with their comparison, Ron's descriptions of comparisons like that are vital if you're to really appreciate what you are (and are not) seeing. Of course, some projection sets get left out entirely and at least Ron got included in their analysis.  I like to think that we're one of the prominent forecasting sites on the Internet but we don't even get mentioned in their comparisons.  Of course, I'm glad that we're not because, again, we wouldn't be getting any credit for all the players we forecast that no one else does, beyond other flaws in the comparison that Ron highlighted.  My biggest criticism of the OPS method is that it gives too much credit to forecasting ability and not enough to forecasting playing time, which is just as important.

By the way, just on your comment of 70% accuracy, I didn't know that this was the consensus but if it is, I think everyone's overrating their skills. To me, we all go in knowing we're going to miss but we're all aiming for the best (i.e. lowest) margin of error. There are so many evaluation systems introduced after the fact to rate forecasts that it's easy to skew results and I try to avoid that.

As far as claims go, since I don't make it an annual habit to rate other people's forecasts, I can't claim to be more accurate than anyone else. What I can say is that there is a comparison system that I would agree to in advance and that's the one we use every year before the season starts. That is, I'll continue to annually publish the exact standard deviations for every forecasted category, comparing our final projection set before Opening Day to the ultimate outcome. Any player we miss forecasting (such as an eventual September callup) has to be treated as if I forecasted a line of 0's and that can work against our overall standard deviation. Any player whom we forecasted to play and didn't play will be treated as 0's in his ultimate result compared to our forecast, also working against our standard deviation. That way, there's no selective choices of players that are rated for comparison players. Everyone who was projected to play or who did play is being considered. If every site would publish such numbers every year, we'd have a good idea in about five years or so whose system was working and which one wasn't. I don't believe a single season would tell the story and so a trend like this would take quite a while to develop, I expect.  If every site would publish a five year moving average of their exact standard deviations using the harsh treatment for missed players above, I'd be open to this being a standard comparison system.  It doesn't mix categories (such as OPS+SLG) and we might discover which sites are the best at forecasting playing time and which are the best at forecasting performance.  Perhaps one site is better than any other at forecasting stolen bases where another might be great at forecasting wins by pitchers.  With such a model, we'd find this out.

By the way, there is one other key issue that you didn't ask about but is on a related topic where I disagree with Ron, as I occasionally do. Ron alludes to fantasy baseball success not really being as dependent on forecasts as you might think and that winning strategies are the key. While I agree somewhat (which is why I so often talk about some strategies that are separate from forecasts such as my so-called "78% Rule"), I sincerely believe that if I could hand you the end-of-the-season stats from 2005 right now, no matter how good a fantasy baseball player you are, I think you'd have trouble losing if everyone else didn't have that info. I could be wrong but I just wanted to highlight that I don't agree with the implication that forecasts are overrated, though I know Ron wasn't saying that they're not important. I just suspect that I rate them higher in importance than he does. As I say, I love the work Ron Shandler and several others do and anyone who claims to be more accurate than some of the best (including Bill James and Shandler, among others, for example) had better say quite clearly what method they're using to rate their forecasts and they better announce that method before the season they're rating begins.  Of course, there are people who think that forecasting and fantasy baseball strategy are two completely separate entities, which is why I always feel compelled to publish such a lengthy testimonial section at the site, so our readers can speak about how they applied both.

For me, standard deviations are everything and that's why we publish our precise ratings annually. I know this is a lengthy response but your question had so many elements that I really did want to cover as many of them as possible.

As always, we'll be publishing summaries of our standard deviations for 2005 before the 2006 season begins and we'll use the same openly-explained calculation method we use every year.  I remain highly respectful of several of our competitors and Ron remains one of them.  In fact, I don't think either one of us has ever claimed to be more accurate than the other, probably because we're both correctly humbled by the forecasting exercise.  Someone once posted a link to Baseball Notebook on an Internet message forum and said that I "claim to the best" and it simply isn't true.  I never will.  I'll give you the forecasts and the results and let you decide for yourself if they're useful to you.

From issue published February 27, 2005:

Q.  Greg Maddux shows as the 7th most valuable pitcher on my rankings when I run your sheets and I am drafting last this year, thanks to winning the league in 2004.  Last year, Johan Santana was showing as the 4th best with your projections and I took him in the first round so I'm wondering if I should take Maddux in the first round if the other three pitchers are gone when my turn comes up.

A.  This may surprise you but those who drafted Santana in the first round last year, even given our high opinion of him, actually didn't follow our recommend strategy.  Even if a player comes out with a tremendous projected value, you shouldn't take him earlier than you have to and you could have waited until at least the third or fourth round, in most leagues, to pick up Santana.  You might have then been able to pick up a Jason Schmidt or a pitcher like that in the first round and then also added Santana.  I know it seems ridiculous to say now, especially since Santana so far exceeded even my high expectations, but it actually was not the best strategy to take him in the first round even if his projected value seemed to justify it.

This, then, is true of Maddux as well.  No one is going to take him in the first round.  No one is likely going to take him in the first five rounds in a small league.  I see that Maddux's average draft position in 5x5s, according to MockDraftCentral.com is 110.5 right now in mixed 5x5 leagues and the earliest he's been taken is 96th overall.  In a twelve team league, that means you should have no trouble picking him up in the eighth or ninth round pick, if you believe the projection.

Remember that draft strategy does not necessarily mean taking players in order of their projected value.  The best strategy is the one that gets you the maximum possible value on your roster any way possible.  If I project Chris Woodward to make a roster and then somehow win an everyday job only to further surprise by hitting 50 home runs, you still wouldn't need to take him any earlier than the very end of the draft or when you're filling the last spot you have to fill, even given the forecast.  Even if it proved to be exactly right, and I make the ludicrous example to make a point, it still wouldn't be good strategy to take him any earlier than necessary.

The 2005 Maddux forecast justifiably proved to be one of the more unpopular projections but readers who followed this advice not to take him early got "burned" by using their "eighth or ninth round pick" on a missed forecast, which should not be a critical blow to any good fantasy GM.  It is interesting that as bad as the forecast for Maddux proved to be, it was the ERA projection that proved to bad.  In the other three commonly-used pitching categories for starters, we had forecasted Maddux to win 14.5 games, strike out 134 and have a 1.12 WHIP.  Maddux finished with 13 wins, struck out 136 and had a 1.22 WHIP.  I don't hesitate to say it was a miss (and to characterize it as anything else would be futile and inaccurate) but it seems that many falsely recall me projecting a better season than I actually did.  He failed fantasy owners but he failed in one key category (ERA).  Readers who followed the advice given here should not have been "burned" and just for what it's worth, I won every fantasy league in which I owned Greg Maddux and many readers wrote to tell me the same thing about their own experience.

I've actually selected this quote not to comment about Maddux but to make the point I made about Santana a year earlier - Drafting players earlier than you have to is a quick route to defeat.  I made a similar point a week later in reference to Pedro Martinez, which I've selected below.

From issue published February 27, 2005:

Q.  I'm in a 12-man head-to-head league that uses most categories for hitters and pitchers.  I'm a little confused after plugging in the points and spitting them out on Excel.  You have Pedro Martinez head and shoulders the overall points leader.  Is he worthy of a 2nd round pick?  I'm not about to pick him 1st.

A.  I've continued to receive many questions similar to this one and I refer you to my answer in the previous edition (which is also available at the site) in relation to a reader who took Johan Santana in the 1st round based on our forecasts before last year.  I told him that despite how Santana did, it actually wasn't the best strategic move.  You don't want to take a player earlier than necessary.

For what it's worth, MockDraftCentral.com shows that Pedro in mixed leagues is going about 23rd overall, on average, and the earliest he's been taken in recent days is 14th overall.  In NL-only 5x5 leagues, he's averaging going 13th overall .  Your points league will no doubt be different but these numbers sound like at least a second round pick to me if you want to end up with him.

Just on a related note, you said that I "have" Pedro the overall points leader.  Remember that where a player is ranked on your fantasy lists is a function of the categories you use combined with the forecast I've created for him.  If the implication of my forecast is that he comes out as the overall points leader in your league, then it's only because I'm projecting Pedro to have a strong season and my forecast, mixed with your own points criteria, puts him there.  Some readers in points leagues are often surprised at who comes out near the top, even when the forecasts aren't so surprising.

Q.  It's my first year using your projections.  I'm wondering if and when your projections will change for players who are battling for a starting job in spring training.  For example, will you change your Chris Burke projection if Houston announces that he has won the starting job at second base?

A.  We constantly update and revise projections based on events in baseball and it's why we just shifted to "weekly" update mode now rather than in April.  Spring training provides us with so much new information that anything less than a weekly update would make it impossible to track the changes and it's why we even offer an extra update in the middle of the week before Opening Day, something we started doing last year.

Absolutely, if the Astros said that Burke would be the starter at second, his projection would go up and I have to remind you that beyond what the clubs say, we also attempt to project who won't keep that job that they've won in spring training.  If a rookie wins a starting spot in March and then goes 0 for April, he's going to be right back in the minors and therefore, you will often see us forecasting a player to lose a job that he apparently wins in March.  For example, Nick Swisher appears poised to start the season as an everyday player but we're forecasting just 65 games and 159 at bats.  It's not that we're not aware that he's slated to be an everyday player but the projected playing time goes in hand with our forecasted .227 batting average.  If we're wrong about his ability, then this is a clear case where we'll be wrong about his playing time too.  Playing time and performance are inseparable, especially with younger and/or inexperienced players.

Yes, spring training events constantly influence what our forecasts look like.  In the case of Burke, whom you used as an example, I'm actually already projecting him to make the team as at least a reserve infielder and pinch-hitter (241 at bats in the latest forecast set), if not more.

The important point here was again, not only the link between forecasted performance and projected playing time but that we're constantly monitoring situations.  In Swisher's case, I was surprised that Oakland stuck with him as he actually did hit as poorly as projected (.236) and yet Oakland, an organization that usually emphasizes the ability to get on base, kept running him out there even though his decent walk ability didn't off-set his lack of hitting in 2005.  Burke ended up with 318 at bats or 77 at bats more than we had been projecting at the time here.

Q.  I read that Adam Dunn thinks he can steal 25 or 30 bases but you've projected just 10 in the latest set.  What gives?

A.  Every year we get at least five or ten players in spring training who say that they're capable of stealing more often and I promise you that I do weigh their projected talent against their desires.  A year ago, I actually over-estimated Dunn's stolen base ability and the latest projection is based on my belief about both his skills and the way I project him to be used.  I don't doubt that he's capable of stealing more often than he does and this is reflected in his 78% projected success rate but I also don't typically buy what players say they're going to do, even though you can occasionally pick up useful information from a player's comments.  In the Best of Ask David: Volume 1, I answered a similar question when Joe Kerrigan once said that he expected Nomar Garciaparra and Trot Nixon to steal 30 bases each in 2002 and I maintained my low-steal outlook in the pre-season projections.  A fully healthy Garciaparra went on to steal 5 that year and the also-healthy Nixon stole just 4.  I explained in that space how it is that a manager or player comment can result in a bump up to the forecast (as Toronto manager Tim Johnson caused me to do for most of the 1998 Blue Jays team) but Dunn's comments alone haven't convinced me.

I also watch early in the season to see whether the so-called green light is a constant reality and I recall that the year Roger Cedeno stole 66 bases, we had him bumped up to a 60 steal projection by late April but I just can't justify increasing Dunn's steal projection until I see the reality of it on the field.  Granted, it means that we will often miss players who have new abilities but I'd rather err on the side of facts than goals, with rare exceptions.  Dunn's got a 49 home run forecast in the latest edition so I'm certain our readers will be valuing him anyway but no, I can't in good conscience give him more than 10 projected steals right now, a number that is almost double what he stole a year ago (6).

I answer one like this just about every year and no matter how they turn out, they always deserve a spot in the "best of" series as they emphasize that what a manager or player thinks someone can do is not necessarily indicative of actual ability.  Vernon Wells also joined this group of aspiring 30 steal types just after I published this response and finished with just 8 steals, many of them late in the season.  As for Dunn, the aspiring base-stealer fell short of even my low expectations, stealing only 4 in 2005, getting caught twice.

From issue published March 27, 2005:

Q.  I'm curious how sometimes you can have a forecast that is similar to another site but when I use the other site's ranking tools, their dollar values come out differently.

A.  There can be several reasons why this happens.  First, if the forecast for an individual is the same but the overall projected population is different, then the scarcity of a certain statistic has different value.  For example, let's say that there are about 2,600 stolen bases in the current forecast set and you have a 26 stolen base player.  That means the 26 stolen base player has 1% of the projected steals.  Even if another projection set out there projects this same player to get 26 steals,  let's say they only have 1,300 stolen bases projected in the entire set.  That would make this player doubly valuable in the context of that other projection set.  Also, we consider what is likely to be the drafted population based on the size of your league and other sites might look at the player population as a whole.  In a fantasy league made up of just five teams, a .280 average or an ERA of 3.90 can actually be a negative thing in terms of value.  In a league made up of thirty fantasy teams, a .280 average or a 3.90 ERA has value to your team.

Moreover, we all use different valuation methods and for good reasons, we keep our valuation methods somewhat private because we don't want other sites copying them, for good reason.  I make no secret that my valuation methods were inspired by (though slightly different) the book Rotisserie Baseball: Playing for Blood, which was published by Diamond Library back in the 1990s.  I was a member of John Benson's writing staff at the time and John and I have frequently compared notes over the years and I have great respect for his methods and continue to occasionally write for his books.  I have great admiration for the concepts he discussed in that series of books and if you want to have a good sense of how I value players, even though I didn't write that book, that's as good a starting point as any.  Our interactive forms are much more complex than anything I could describe even in a full book (the ranking sheets took us four years to develop and have several thousand lines of code to account for the variations in leagues) but if you're getting different results from other sites for apparently similarly ranked players, it could be because of the way it views the player population or it could be because of the valuation methods.

Finally, a projection could seem similar on the surface but actually might be quite different.  Adding even five steals to a forecast can significantly boost the value of an otherwise identical forecast, especially in a small fantasy league made up of ten or fewer teams.  Also, our interactive forms use projections that go to one full decimal place, even though the numbers appear rounded on the screen.  This can impact the values as our forms discern the difference between a player projected to hit 1.6 home runs and a player projected to hit 2.4 home runs even though on the rounded totals they're identical in terms of how they would seem to impact home runs (2).

From issue published April 6, 2005:

Q.  Does Dmitri Young's three home run performance on Opening Day qualify as being outside the margin of error and can we now safely say he's going to have a huge year?

A.  If Young were the only player we were considering in advance and I didn't care about any player, yes it would be outside of the margin of error but when we're considering every player in baseball, it is entirely predictable about every fifteen or twenty years that a player, of unknown identity in our complete set of players, will debut with three home runs in an Opening Day game.

Two other players in history homered three times on Opening Day.  George Bell, coming off an AL MVP in 1987, homered three times to start the 1988 season and looked to be every bit en route to repeating or even exceeding his 47 home run season of 1987.  He finished with just 24 home runs in 1988 despite getting 614 at bats.  Karl "Tuffy" Rhodes was the only other to hit three home runs on Opening Day, in his case to start the 1994 season against Dwight Gooden.  He went on to finish with 8 home runs in 269 at bats for the year.

No, as we're considering the entire population and can't retroactively decide that Dmitri Young was the only player of interest to us, his performance is completely within the margin of error and he deserves no modification to his forecast based on one performance.

Young finished with 21 home runs, including those three from Opening Day, right about where he should have.

From issue published April 13, 2005:

Q.  I stuck to your strategy and got these three guys for my outfield: Abreu, Andruw Jones and Vernon Wells.  You have projected very strong years for all and as of Friday night these have combined to go 9 for 49 (.184).  Not exactly a hot start.  I hope they pick it up soon.  Are they all slow starters?

A.  A lot has changed even since you wrote me this question and I'm publishing it to make a point but first let me answer the question re: slow starting.  Though I recognize that there are players who are "true" slow starters, I'm not a huge believer that there are many players who are true slow starters.  Many are considered to be traditional slow starters because they usually had one or two bad years in which they had tremendously terrible starts (like going 1 for 40) and then those stats get combined with other April stats from other years which are actually fairly reasonable and it makes it look like the player is a slow starter when in fact they had one bad start to one season.  Beyond that, the nature of slumps and streaks is that by chance alone a player can start poorly or well and it may have little to do with his skills.

Your question, though, is quite representative of a theme of messages submitted to this column and they involve players who, in the first week of the season, had really great or really bad starts to the season.  Whether it be hitters who started by going 0 for the week or pitchers who were hammered in their first start or two (John Smoltz was a popular topic among resigned fantasy owners who watched his first start and he quickly redeemed himself by striking out 15 batters in start number two), everyone puts too much emphasis on the first week or two of the season.  Even with my column last week, people keep writing me about Greg Maddux, who has just two starts in the books, wondering if I'm already ready to downgrade him.

It takes about 20-30 games for a hitter to achieve a large enough sample size for us to do a reliable revisiting of his skills.  It takes about six or seven starts, at least, for a starting pitcher, to achieve enough of a sample and for bench players and relievers, it can take until the middle of the season.  I get people who tell me "that's not quick enough to make decisions" but it's a reality of forecasting that you can't overreact to the hot or cold start because you will quickly trade down your team to non-value entirely.  One of the most common ways to lose is to have a plan and then abandon it too early before you have a chance to find out if it would have worked.

A good example of this can be found in the three players you mentioned.  As you say, as late as Friday, they had combined to go .184.  As of the moment I finalize this issue, Abreu is now hitting over .300 and has since had a nice three stolen base game, Andruw Jones is hitting .333 and Wells is the only one of the three who hasn't come around yet and even at that, the sample size is still too small for us to know whether any of these three have different abilities than we forecasted, in either a good or bad way.

Players in the long run will perform according to their abilities but in the short run, particularly over a week or two, just about anything is possible unless the player is completely overmatched.  Don't be too quick to react because you'll end up trading your strengths for players who started well and performed above their abilities.

Every April, particularly in the first week or two of the season, much focus is on how players have done in their first 50 at bats.  It all means little in terms of the forecast, unless it's a rookie or non-established player struggling early (who can then lose his job) and Abreu, Jones and Wells all ended up with solid years.  Jones and Wells would come up frequently as their struggles continued into May, as we'll see later.  Another good example was in the same issue below.

Q.  Is Mariano Rivera declining now?  I'm wondering if he will be one of those closers who ends up not being valuable to my fantasy team this year.

A.  Mike Fitzpatrick of The Associated Press wrote an interesting article last week posing the rhetorical question "Rivera's Dominance Done?" in the headline that I assume was the reason I had a lot of readers writing to ask if I thought the same.  The article cited examples of how Rivera had blown three saves in last year's playoffs and talked about his problems against Boston.

I don't think this is the year of an instant decline and citing examples of blown saves against one team, a good-hitting one at that and the best-scoring team in the league last year, proves only that he hasn't been effective against them.   He's set such a high standard that no doubt everyone expects him to be as perfect as he always has been and that's probably too much to ask.

One of the reasons I don't think he is in decline is because declines, while they can be sharp, tend not to be so cliff-like with players who have recently been at his level unless a career-ending injury strikes.  Mariano Rivera has posted a sub-2 ERA each of the past two years and for me to expect a decline this year, I would say that his ERA for 2005 would have to finish above 4.00.  That would qualify as a sharp decline for me even though I may even be holding him to too high a standard.  I did some research on this and found that there were an amazing 63 pitchers since 1900 who were coming off two consecutive seasons in which they had pitched at least 50 innings and posted ERAs of better than 2.00.  Of those 63 pitchers, the combined ERA in season three was a microscopic 2.12 with more than 14,000 innings logged by this group in year three.  Amazingly, you have to reach back all the way to 1918 to find the one and only pitcher out of the sixty-three who failed to have an ERA of below 4.00 in year three and that was Ferdie Schupp of the New York Giants who pitched just 10 games after two stellar seasons in 1916 and 1917.  Not a single pitcher in history since has pitched at least two consecutive years of sixty innings with sub-2 ERAs and had an ERA of worse than 4.00 in year three.  The worst performance other than Schupp's was Tug McGraw's 3.87 ERA in 1973.  A remarkable 51 of the pitchers (or 81%) posted an ERA of lower than 3.00 in year number three.

In other words, if Rivera is declining, which is entirely possible at his age, I don't expect the decline to be instant and he should still be valuable to you and your fantasy team.

Rivera was actually an extremely popular topic in April when he struggled and readers were sending me all these media links to various columns speculating on how he was through.  Of course, Rivera's reported decline ended up in him finishing as the runner-up for the Cy Young Award.  Several readers thanked me for the research I did here in my response and so, it makes it into this collection.

Q.  What is your opinion of Huston Street, both for this year and as a keeper.

A.  I like this guy for both purposes and my current forecast is for a 3.05 ERA the rest of the way, even though his moderate playing time forecast the rest of the way (29 games) represents the high degree of uncertainty you must account for with such inexperienced players.  He's good enough to be a closer and has Octavio Dotel in his way but from a keeper perspective, you have to remember that one of Billy Beane's excellent tricks is to create established closers, boost their value to the market that values saves more than stuff (though Dotel certainly has the stuff too), and then trade them out of town to bolster other areas of his roster.  Therefore, I could see Dotel being an ex-Oakland A by 2006 or 2007 and Street being one of the strongest contenders to replace him as the team stopper.  Besides that, he's one of the primary candidates to replace Dotel in the event of a 2005 injury and I don't think he'll hurt you to have on your roster while you're waiting for him to get a prime time opportunity.  Get him if you can and you better hurry.

Regardless of how well Street ended up performing, the advice here applies annually and that is that you have to watch Billy Beane when it comes to closers because he correctly recognizes that the market tends to overvalue them based on their experience in the role.  It wouldn't surprise me to see Beane's friend J.P. Ricciardi take advantage of the same by trading Miguel Batista this off-season while his value as an "experienced" closer is apparent.  I could never have anticipated the Dotel injury, obviously, but readers who followed the advice here were rewarded with a lucky immediate result with Street that I hadn't even projected when I recommended picking him up.

Q.  I'm surprised you've forecasted a poor WHIP for Brian Bruney.  In the minors, he allowed only a half a home run per nine innings pitched, held opponents to a .177 batting average and didn't walk many.  I'm very high on him.  Why aren't you?

A.  This may be a simple case where you're going to have to be content to disagree with me but I think my forecast is for a pretty good pitcher so it's not as if I'm not high on him, as you say.  If the WHIP is your primary focus, as it sounds like it is, then the home run rate won't play much of a factor there as home runs represent only a small portion of the hits any pitcher allows and a single is just as bad as a home run from a WHIP perspective.

You quoted a lot of minor league numbers there but I think they deserve to be put in context.  I agree with you on his excellent ability to prevent home runs and my forecast reflects that as I'm forecasting just 0.55 home runs allowed per nine inning in the majors, which will help him to be successful.  I'm also forecasting that he'll allow fewer than one hit per inning and so your quoting of the batting average is also in line with my forecast.  When you say he didn't walk many, that's where I absolutely disagree with you.  In 38 minor league innings last year, he walked 20 batters and since 2002 in the minors, he's walked 55 batters in 113.2 innings.  That is an example of a pitcher who does walk many and if you pro-rated that to a recognizable 200 inning performance, such as what a starter would have, that's 97 walks.  This is not even accounting for Double-A and Triple-A hitters being far less selective than big league hitters and my forecast of his wildness accounts for the difference between the quality of competition of the minors to the majors, meaning he should walk major league hitters at least at a slightly higher rate than he did in the minors.  Even un-translated, that minor league walk rater per inning is almost exactly the same rate as the 2004 version of Miguel Batista, who led the American League in walks last year.  So, I strongly disagree with you that he "didn't walk many" in the minors and that's the crux of our disagreement on his projected WHIP, which seems to be the element of the forecast in which you're most interested.

The important point here was that we have to look at minor league statistics in their entirety rather than pick out the ones that suit our high (or low) opinion of a player and especially when it comes to minor league relievers, the walks often don't jump out on the page because they pitch so few innings but they can indicate future problems.  Bruney finished 2005 with a disastrous 7.43 ERA, 1.97 WHIP and 46 innings that hurt every fantasy team that otherwise would have benefited from his 12 saves.  His performance was bad enough that even the saves didn't offset it.

From issue published May 4, 2005:

Q.  Is Vernon Wells really this bad?

A.  Almost exactly a year ago, I responded to two similar questions about both Derek Jeter and Ichiro and for different reasons, explained why each should be expected to rebound.  In Wells' case, it's actually more of a certainty as you'll see shortly.  I should warn that this section is for the more mathematically inclined who want to know how it is that I really come to a forecast and just as importantly, what will cause me to modify it when a player starts well outside what we would have imagined.  A couple of the answers this issue are a bit lengthier than usual because I want readers to really understand the thought process that goes into the forecasts and why it is that some players don't get downgraded when readers expect them to be.

Wells' contact rate (I encourage the reader who might not understand the concept of contact rate to read about it here) last year was 85% and his career rate coming into this season for his career, over 2,008 career at bats, was 86%.  In fact, in his three full seasons in the majors, his contact rates have been 86%, 88% and 85% so it's pretty obvious that there's consistency there and a true ability, especially considering he hasn't even turned twenty-seven years old yet.  Through games played this year on May 3rd, Wells' contact rate was 85%, right where it should be based on his ability.  That may surprise readers as they may have felt like he was striking out more often than normal in 2005, which he is not.

Now while some argue that he's always a slow starter, and historically that's been true, I don't buy that theory about his start either.  The way to break down his performance is to examine it on a "per ball in play" basis just as I described in that column I alluded to a moment ago that explained the contact rate.  His singles per ball in play rate this year is 15.05%.  His doubles per ball in play rate is 3.23%.  He has one triple to give him a 1.08% and his home run per ball in play rate is 4.30%.

Let's look at the extra base aspects of his performance.  On Opening Day, the forecast I published for Wells had the following projected rates for singles, doubles, triples and home runs per ball in play respectively: 21.13%, 7.98%, 0.67% and 5.45%.  Looking at the home runs, which is always the most interesting category, of the 93 balls he put in play so far in 2005, our forecast from Opening Day said that he should have hit 5.45% of those for home runs or 5 home runs.  He has 4 home runs so if you're concerned about that when his forecast said he should have had 5 by this time, then you really are being a stickler for accuracy as I'll settle for a margin of error of one home run per month for every player, if it was offered.  He compensates for this by having hit his first triple, which the forecasts wouldn't have expected to have happened yet.  The problem so far completely relates to doubles and singles.

The doubles that should be expected if we could have known how often he would put the ball in play (and we were lucky enough to be right about the balls in play as he's putting it in play almost exactly as projected) should be 7 doubles and he has just 3.  For singles, we would have expected between 19 and 20 singles and he has just 14.

Just to show you how it is that I decide whether to revise a forecast, I have to look at the new apparent elements of a performance and weigh them against the evidence that drew me to the original forecast.  Wells' double per ball in play rate for his career coming into 2005 was 7.53%.  His rates for the three full seasons he's played in the majors were 6.50%, 8.19% and 7.51% for 2002, 2003 and 2004 respectively.  So, let's go ahead and say I was completely wrong (which I'm not really prepared to say but we'll do it for hypothetical purposes to make a point).  Let's say that both 2003 and 2004 were complete flukes and his real doubles ability is 6.50%, the low rate he hasn't shown since 2002.  If that's the case, he should still have 6 doubles right now instead of 3 and that's if we accept that his worst doubles per ball in play year in the majors represented his real ability, which would be a big mistake.

But the singles are the problem in his performance and that's the easiest one to show is simply a result of bad luck and not, as some would say, his propensity for bad starts.  His singles rate per ball he's put in play this year, as I said already, is currently at 15.05%.  What you need to know that is in the history of baseball, or at least every hitting performance since 1910, of players who ended up with at least 500 at bats in a season, there have been just 10 players who ever had a singles rate per ball in play so low and there hasn't been one since 1995.  The reason?  It takes amazing bad luck to have a singles rate so low.  Any player can be almost certain that his singles rate will be better than Wells has been so far because it's actually difficult to maintain such a low singles rate.  By luck alone, just about any player who plays big league baseball will get more singles per ball in play in the long run than that rate even if he isn't a good hitter and that we've averaged about one player every ten years with such a low rate is evidence of this.  Here's the better news:  These ten players I mentioned were all failing to hit singles because they were all hitting plenty of home runs - every one of them.  The lowest total accumulated by any of the ten in home runs was Roy Sievers and his 24 home runs in 1954, which put him among the league leaders in home runs that year in a year when only Larry Doby hit 30 home runs or more.  The average number of home runs hit by these players was 37.4 and this list of ten includes the seasons of Roger Maris in 1961 (and his then record-setting 61 home runs) and Albert Belle's 50 home run season of 1995.  These players were trying to hit the ball out of the park and thus, when they failed to do so, their upper-cut swings caused them to achieve surprisingly low singles per ball in play rates.

Among players with at least 500 at bats in 2004, the lowest singles per ball in play recorded was by Carlos Beltran (yes really) with a 15.46% rate.  That's because he was doing well at swinging for the fences.  He was followed by Tony Batista, Vinny Castilla, Alex Gonzalez, Brad Wilkerson and Albert Pujols, four of which hit at least 30 home runs.

So getting back to Wells, there's no way his low current singles rate per ball in play will last.  I said the exact same thing about Ichiro almost a year ago to this week, when he was struggling with a slow April in the books and several readers were doubting whether to keep him.  Just like we did with the doubles, let's consider Wells' singles rate so far in his three full seasons.  His singles rate for his career is 20.79%.  In his first three full seasons in the majors, he has rates of 20.27%, 21.40% and 19.21%.

If you took the worst singles rate of his career (from 2004), the worst doubles rate of his career (from 2002) and the triple and home run rates he's even showed so far this season, he should rightfully have 29 hits by now in his first 109 at bats or a .266 average and that's being as pessimistic as you can possibly be on the singles and doubles rate.  In other words, Wells is one of those rare players where I feel entirely comfortable telling you that there is no way this slump can last much longer, even though with his April performance mixed in, it's going to look like it lasted longer because everything he does from now on is mixed with that slow start in the year-to-date stats.

Like the Melvin Mora explanation I gave in January, I couldn't possibly publish as much detail behind a forecast as I did here for Vernon Wells on May 4th but it became necessary as Wells and Andruw Jones were so dominating my email in box that I was forced into a public, detailed response to why I maintained a high forecast for him the rest of the way.  Wells did right himself and finished with a .269 average, 28 home runs and 97 RBI.  Of course, the .269 average included the disastrous start to the season and from this point forward, Wells performed almost exactly as projected.

Q.  It's obvious you were wrong about A-Rod having a comeback year so when will you admit you were wrong?  I'm going to bench him today for at least a week or two until he starts showing he can produce like a first round pick - which is what he was for me based on your projections.

A.  Wow.  This email was sent for consideration to this column on the morning of April 24th, hours after our projection update and a Sunday when many leagues allow their weekly moves.  The reason I say "wow" is because this is my first chance to respond to the question in public as this is the first "Ask David" to run since the question was submitted.

That means that, assuming you benched him as you threatened to do here, the week you missed is one that started with A-Rod enjoying a tremendous 3 home run, 10 RBI game.  I will neither say I was right nor wrong about Rodriguez so early in the season but his performance in a single week, if not a single day, emphatically demonstrated how quickly a season can turn around at this time of year.  When you wrote me with this question, he was hitting .269 with 4 home runs and 14 RBI, which was still, I thought, decent production to start the season.  Now, just over a week later, he's hitting .295 on the year with 9 home runs and he's tied for the league-lead with 27 RBI (I've projected him to lead the league in RBI this year but we'll see if that holds up).

Just as I couldn't answer you when you wrote me, I can't now either.  Because of the nature of forecasting, I can virtually guarantee you that he won't finish with the exact totals I've projected in every category.  If that makes me "wrong" by your estimation, then you have my admission already.  If you were expecting him to finish with 14 RBI on the year, then there's nothing more to say either.

The example here was that impulsive decisions can really hurt a fantasy team because sometimes great performances can come in clusters and if you miss out on a cluster, you can't always make up for it.  A-Rod didn't end up leading the league in RBI, as I had been projecting, but finished fourth and certainly didn't disappoint readers who went with his pre-2005 forecast.

From issue published May 11, 2005:

Q.  Should I trade Hideki Matsui now?  He has not had a home run in 20 games now and his average and power is dropping fast.  Are you going to downgrade him?  Lately, leaving him in my lineup makes as much sense as eating soup with a fork.

A.  When a player struggles, as Matsui has, that's the absolute worst time to trade him because trading partners will only give you perceived recent value and you'll be constantly "trading down" your team into nothingness.  If you really believe, as I don't, that Matsui's struggles represent a real and instant decline in skill, then of course the trade is the smart thing to do.  No, I'm not about to downgrade him more than he deserves.  Compared to the Opening Day forecast, there has been a slight downgrade but I have no doubt that he's a much better player than his year-to-date stats imply.

Andruw Jones hit 1 home run in his first 17 games this year and I was receiving many questions with many people ready to dump him too.  He's since hit 5 home runs in 16 games.  See the question from last week's issue from the reader who wrote to tell me he was benching A-Rod for further evidence of just how suddenly a player can arrive for the year with stats you could easily miss if you abandon your plan.  Things can turn around quickly and Matsui can and will set things right, assuming he's not injured.  Regardless, you won't get fair value for him right now.

Again, the danger of readers reacting to early stats was underscored here.  Hideki Matsui got hot about a week after this was published and never turned back, finishing the year with a .305 average, 23 home runs, 108 runs scored and 116 RBI.

From issue published May 18, 2005:

Rather than re-publish the lengthy excerpt, I'll provide a link to it here and explain.  By mid-may, after addressing Vernon Wells a couple of weeks earlier, virtually 70% of the questions to my in box at the time mentioned the same name over and over and that was Andruw Jones.  Readers know that I had projected a career-year for Jones in power and I was maintaining that outlook into mid-May even as he struggled to a sub-.200 average.  The topic became so popular, particularly with irate readers (one even demanded his money back for his subscription, citing that it was "only because (I) continue(d) to project Jones to hit a career-high in home runs) that I felt a full column had to be dedicated to Jones, Wells, Maddux, et al.  You can read that full-length column which will soon find its way to the archives at the site as I entitled it "On Fantasy Baseball Trading" but it was entirely motivated by questions to my Ask David in box.  You can read it by clicking here if you're interested.

From issue published June 1 2005:

Q. Damon Hollins seems like a good, young player but your projection is pretty tough on him.  He's leading all rookies in home runs and is starting every single day.

A.  I'm probably going to give Hollins an upgrade in projected playing time in this upcoming weekend's projections but in terms of his ability, I think I'm being pretty fair.  If you pro-rate the projection over a typical 550 at bat full season, he would be a .259 hitter with 19 home runs and 8 stolen bases.  He hit 20 home runs last year at Triple-A Richmond and the previous two years had hit 23 home runs in about 800 at bats combined those two years.

You say he's playing every day but I don't agree that this is the case anymore.  He did start just about every game from May 2 - May 15 but his playing time recently is becoming more sporadic.  In the past seven days, he's started only four of Tampa Bay's games and was on the bench last night against Oakland's Danny Haren (Hollins did come into the game to replace Aubrey Huff late and ended up without an at bat).  As I said though, I do recognize the need for an upgrade in projected playing time, which you'll see in our next projection issue.

Oh, and you referred to Hollins as a good, "young" player.  I need to correct you on that and it's easy to make the mistake of confusing a lack of experience with youth.  Hollins, who made his major league debut with Atlanta last year, actually turns thirty-one years old in the next two weeks.  He's a 13+ year minor league veteran, just in case you were wondering, and certainly does not qualify as a prospect no matter how he hits.

The lesson here, as I first read in Bill James, is that we must never confuse a player's inexperience with youth.

From issue published June 8 2005:

Q.  We're having an argument in our league.  Do you think Chris Shelton of Detroit should be eligible at catcher?

A.  Absolutely not.  I can understand that if someone (perhaps you are on this side) argues that minor league positions should be counted when determining eligibility.  Otherwise, players like Clint Barmes would have come into 2005 only qualifying for the dreaded utility spot but let's go ahead and say that minor league position does count.  Shelton caught just 6 games in the minors in 2004, playing DH 19 times and first base 17 times.  He also played 1 game in the outfield.  This season in the minors, Shelton caught 3 games for Triple-A Toledo but played first base 31 times and DH 14 times.  Just an interesting trivial note on this is that Shelton has actually never had a minor league season when he played catcher more than any other position.  The year he had the most games at catcher (when he played 38 games there at Single-A and Double-A combined in 2003), he played first base 43 games.

In other words, Shelton has never had catching as his primary position, even in any minor league season and with just 9 games at catcher in the minors since the start of 2004, there's no way any league should be qualifying him at catcher until he qualifies in the majors there, which he won't anytime soon.  If you're going to use minor league positions to qualify a player, Shelton's primary position in the minors last year was DH and this year it was first base.

I've picked this one as an example of a sin that many fantasy leagues are guilty of and that is letting players qualify at a position long after they haven't played it professionally.  Watch for this in your league and be quick to contest it because opponents can often fill their weakest spots with players who shouldn't be qualifying there.  Of course, one simple solution is to simply make minor leaguers have to come up through the utility/DH spot.  It eliminates most arguments but loses a bit of realism in the process.

Q.  I'm curious about your projection of only 54 more innings for Glendon Rusch.  According to your projection, which is for a solid 3.61 ERA the rest of the way, he would take on great value if we determined that he will get 110 innings instead of 54 the rest of the way.  I've been offered Rusch for Pettitte and I'm considering it.

A.  Rusch is projected to make about 5 more starts the rest of the way against about 25 relief appearances.  If you're wondering why there's so few relief appearances, remember that when a pitcher starts, he's unavailable for about four or five days meaning that if Rusch gets five starts, which you obviously disagree with, it means that we're talking about projecting how he will be used in relief in the time he's not a starter.  The Cubs had about 110 games left when the last projection set went to press.  Figure that if he starts 5 games in those 110, there's about 20-25 days when he isn't available to pitch in relief, which leaves him pitching in 25 games out of about 85 days or about once every three days.

So, let's talk about the starts.  I may be wrong but I really do believe Rusch will end up back in the bullpen, even as good as he's looked.  His track record is very heavily working against him here, not only in my eyes but in what I expect will be the view of the Cubs' front office (coming into this season Rusch had a career ERA of almost 5.00 in almost 1,200 big league innings).  The key here is for you to understand that if he does stay in the rotation, it's not going to be because he has a 3.61 ERA.  As I often say, when you see a pitcher projected to have a good ERA with plenty of games in relief, it can often be because I expect him to pitch better as a reliever than I do as a starter.  If he stays in the rotation, his ERA is going to be much higher than that because relievers typically have an easier time posting a good ERA than starters because they don't have to pitch full innings.

So, if you think I'm wrong about how much Rusch will pitch, then by all means upgrade the number of innings but if you do so, don't expect him to have an ERA below 4.00 then the rest of the way.  You may very well be right about him but I still believe that before long, despite how Rusch has pitched, he's not going to keep up his great start to the season, even though he won't kill your fantasy team if he does stick in the rotation.  I expect that by the time Prior is healthy, the rotation will be a solid Prior, Wood, Zambrano and Maddux with all of Sergio Mitre, John Koronka and Rusch fighting for starts out of that fifth spot.

It's important to remember that often, a pitcher will be projected to have a decent ERA because he's projected to be a reliever.  Readers often justifiably ask why such a pitcher wouldn't keep his spot in the rotation if he had that good an ERA and my response is always that if he stays in the rotation, I wouldn't project the ERA to be as good, naturally.  It wouldn't have changed the point but Rusch did end up back in the bullpen, though injuries forced him back to the rotation in September, and he finished the year with 46 appearances, 19 of them starts.  He pitched much more poorly than I had expected the rest of the way.

Q.  You still have Derrek Lee projected at 28 HR, 70 RBI and a .289 average.  Why do you keep his numbers so low seeing that he will probably shatter them by the All Star Break.  He has a .380 average, 17 HR and 50 RBI only 187 AB into the season.

A.  Member support responded to this reader directly but I suspected that the same confusion might exist out there so I wanted to clarify.  The forecast you quote for Derrek Lee, while correct, is his forecast for the remainder of the year.  Our forecast sets always project only the portion of the season that hasn't been played because that's what's useful - it does you no good to know what a player has done or at least it cannot possibly not have any new value to you except in analyzing ability and playing time.

Just while we're on the topic of Derrek Lee, in fact I forecasted a career-high 39 home runs for him on Opening Day and at least early, he's turning out to be one of the forecasts that has most benefited readers because he had never hit more than 32 in a season before this year and had consistently hit between 28-32 home runs four of the last five years.  Lee, at least early, is turning out to be this year's version of Adrian Beltre for us.  In other words, he won't be the best forecast because he's on a pace to exceed our original expectations (remember that performing much better than we expect is also a miss, contrary to what some people seem to expect) but like with Beltre a year ago, we did project a career-high that pushed Lee much higher on our lists than common perception would have valued him.  Here's hoping he keeps it up for all those Derrek Lee owners out there.

This note, which is necessary about once or twice a year, was a reminder to all readers that at any point in the season, the forecast is for what I expect the player to do for the remainder of the season only.  From this point forward, Lee hit .315 with 29 more home runs and 57 more RBI.

From issue published June 15, 2005:

Q.  I believe that players do better in their final year before free agency.  Adrian Beltre is a great example of this.  I'm wondering what your take on this is and just how much better they do on average?  It seems that time and again a player who is eligible for free agency has a career year.

A.  Back before we were on the Internet and up until about 2002, I wrote an article about this topic every single year until I finally tired of proving the same point over and over and the results of my analysis every year were the same, dating back to the first version of the analysis back in 1995.

Basically, every analysis I did on the topic (from 1995 to 2002, I revisited it every year except one), came out with the same conclusion and that was that there was no statistically significant difference between how the average player performs in his final year prior to free agency and in other years, though there is no doubt always heavy anecdotal evidence to support this with occasional players but not enough to offset the full volume of data that defies the theory.  There could be any number of reasons, not the least of which is that players are really trying hard even when they're not potential free agents.  Baseball remains, after all, a competitive sport and players want to stay in the majors.

If you want to read the most recent analysis I did on this subject, you can see that article by reading "The Potential Free Agent System" at the website.  I don't have anything new to add and as I say, after doing the same analysis seven out of eight years and consistently yielding the same results, I decided that there was nothing new I could contribute to the topic beyond what I had already done.

This is another question I get every year and by including it in this compilation, I'm hoping for all but the newest of readers, my opinion on this is clear as it hasn't changed.

From issue published July 20, 2005:

Q.  About three weeks ago I was trying to get Mariano Rivera from another team and he told me he wasn't available.  Then, just yesterday, this guy turns around and trades Rivera to his friend for a deal that was less than what I had been offering him!  Do you think fantasy leagues should regulate things so that a team has to take the best offer available rather than trading with their friends?

This question is similar in concern to the one I just answered above, though an obviously different issue.  Both go to league rules, though, and that's why I'd recommend that you check out that old essay of mine I mentioned above (it's available in the archives section at our main page as well).

Here's the problem I have with regulating as you suggest.  Who decides what the best offer available is?  In the example I quoted in my old essay, I talked about how the league was sure I was getting the raw end of a deal that I felt confident about.  Sometimes people say things are "obvious" but I argue that they're not always as obvious and that when it comes to rating baseball value, there are few self-evident aspects of a player's value.  Before I empathize with your frustration (and believe me, I do), consider that it's possible that the deal the opponent got really was the one he thought was the best for his team.

I think the real element of your concern here, and understandably so, is that you don't like that he dealt with his friend, helped his friend's team, rather than yours.  My argument is that it's one problem with the fantasy world as opposed to the real world.  It's not too common, though it does occasionally happen, that in real baseball, people go into the business being friends with other owners in advance.  It can happen and there are many "baseball families" that span several organizations but with all the players involved, owners, GMs, etc, and with the emphasis on the bottom line, teams still try to look for the best possible value as opposed to helping their friends out, though there are clearly common trading partners in baseball because of association or former association of the individuals involved.

Sadly, the best advice I can offer is that you find a way to make yourself the trading partner of choice by making solid offers that appear as irresistible as possible.  I was in a league about ten years ago, one I've long since left, where there was one team owner who wouldn't deal with me.  He complained constantly that no matter what I offered him, the deal always went wrong for him later and he always figured, incorrectly so, that I was using inside information against him - like I had a crystal ball that knew when a player was going to break his leg or something - which I didn't.  Ultimately, I figured out how to handle this fellow.  He kept rebuking my offers and trading with his favorite trading partner.  So, what I'd do is go to the team he traded with and aggressively go after the player I wanted after the fact.  Often, I'd end up giving this second team more for the same player he wouldn't trade to me.  It even hit a point where he tried to have trading partners agree not to trade with me because it became a pride thing for him.

Anyway, the short story here is that I've never put emphasis on the individual player.  It's the type of player you're trying to acquire that's important and not the player and his name.  There are rare exceptions in baseball but what you should be looking for is saves from a good quality reliever and not specifically Mariano Rivera.  Once you become attached to the individual player you're after as opposed to the individual traits you're after, you're already locked in.  If he traded Rivera to his friend and his friend won't talk, then move on to the next team.  There is a point at which it's time to leave a league because the bonds between opposing teams become so tight that it leaves you no room to maneuver but in general, I'd rather just pay them back by finishing first and that's the best way to resolve your feelings.  There are saves out there and there are good closers.  Go for the attributes you need as opposed to the names and you'll quickly put your justifiable frustration behind you.

From issue published August 3, 2005:

Q.  I have the #1 waiver priority in our league.  I am awaiting Delmon Young but have recently read that he may not be brought up to the major leagues this year.  In order to pick up a minor leaguer, he has to be elevated to the majors in my league.  Should I grab Felix Hernandez when he gets brought up or do you see Delmon Young getting a cup of coffee in September?  I would rather have Young but don't want to end up getting a sub-par prospect or nothing at all waiting for him.  I already have Scott Baker and would like to wait for Young but am starting to wonder if he will be brought up at all this year.

A.  In the answer to the previous question above, I alluded to so-called "percentage plays" and you're in an obvious one here.  Felix Hernandez is about to be called up to the majors this week and in fact, he's expected to be activated for tomorrow's game.  You simply have to take him, even if you want Young.  Here's my thinking.

I'm going to get a bit mathematical here and readers will have to forgive me for that but I really want you to be a good fantasy baseball player and at the same time, I want you to play this the way you should, based on your beliefs as you've stated them here.  I'm incorporating an oversimplified version of game theory here but you'll see why it can be valuable to approach it from this perspective if you can put up with (what I think is) very simple math.

Let's say that Young has a 60% chance of getting called up in September.  This might be overestimating his chances but I certainly don't think I'm underestimating his chances of a recall.  The Tampa Bay Tribune even had a story published just days ago (which you can read here) that implied that Delmon Young may not even start 2007 with the Rays so to me, a 60% chance of a recall in September looks like the high end of things.

So, let's say we go with that 60% as a best case chance of a recall in September.  Now, barring a missed flight, stubbed toe or spider attack, Felix Hernandez gets called up tomorrow so we assign him a 100% chance of a recall (just to round things for the sake of example - it really is more of a 99% chance even when a team says a player is getting called up).

Now, that means that if we played the season a billion times, if you claim Hernandez every time, you end up with him every time and you end up with Young no times.  However, if you skip Hernandez, you get Young 60% of the time and no one near as valuable 40% of the time.  In other words, you basically average getting 60% of the value of a Delmon Young like prospect, if you catch my meaning.  So, the only way you should not claim Hernandez is if you believe that Young is worth about 167% of the value, prospect-wise, as Hernandez.  Here's why:  If you average 60% of Delmon Young in your billion seasons played and he's worth 167% of Hernandez, you end up actually breaking even (i.e. 60% of 167 = 100).  Again, that means that if you believe that Delmon Young has a 60% chance of a recall, then you take 1 / .6 = 167% and that's the value Young needs to have, in your view (not mine) compared to Hernandez to justify not claiming Hernandez.

Let's say Young only has a 30% chance of a recall this year.  That means that 1 / .3 = 333% which means that Delmon Young, in terms of his value as a prospect, would have to be more than three times more valuable than the way you view Hernandez to justify not claiming Hernandez and waiting on a claim for Young.

Much like people who play poker, you'll probably curse my explanation if Young gets the call in September (poker players call this "being burned by the river") but the right play isn't based on the outcome but by playing correctly over the long run.

So, I say the high end chance of a call-up of Young is actually more around 50% so he'd have to be twice as valuable, in your eyes, as Hernandez, to make saving your #1 waiver pick justifiable as a smart percentage play.

By the way, one possibility you might not have considered is that Hernandez has such an incredible reputation as a prospect that you might even be able to trade him for Young if someone else later claims Young.  It's just an idea.

Anyway, for me, I'd be claiming Hernandez this week and you should too.  Not that you should ever make your decisions based on emotions but of the two negative scenarios you can imagine with negative scenario "A" being that you claim Hernandez and Young later gets recalled and negative scenario "B" being that you don't claim Hernandez and Young never gets recalled, which one do you think would be bothering you more during the winter?  I assume my opinion here is crystal clear.

Even if Young had ended up getting called up (he didn't), my advice here would have been the same.  Many readers commented on this and thanked me for the "percentage play" explanation.  I later actually explained why despite my belief that he probably wouldn't get a call-up, he deserved a forecast heading into September.  That excerpt appears later in this compilation.

From issue published August 17, 2005:

Q.  Eric Chavez's BB/K ratio is looking pretty bad compared to last year's, when it was almost 1/1.  What do you attribute this to?  Is he getting pitched around more than last year, is he in a slump or was last year a fluke?  What do you think of his future?

A.  A few weeks back, a reader asked me a question about how many of the 35+ HR hitters in our Opening Day projected set were projected to achieve career highs in home runs this year.  I commented then that of the seven such players in our Opening Day set, only Chavez looked likely to fall short of the career-setting pace.  In other words, my expectations for him this season were quite high and he looks like the miss of that group, barring a historic finish to the season.

Regular readers know that I don't get too wrapped up in the popular ratios that other sites do (the similar common one being K/BB ratio for pitchers), even though I consider everything when making a forecast.  To me, the problem with ratios is that sometimes one of the two or more categories being compared can be examined on its own to give you the answer to the question.

You're right that Chavez's BB/K ratio this year is lower than last year, standing at 2.4 through games completed on August 16th.  Last year, you correctly note that his ratio was a strong 1.04 rate.

But let's look at the actual performances to answer your question, separating walks from strikeouts.  Chavez's contact rate this year, again through August 16th, was 77.9%.  His rate in 2004 was 79.2%.  So, he's within about 1.3% of last year's contact rate which is easily within any margin of error for even medium confidence levels.  That means that the strikeout column is not the concern and were we to focus on the ratio of BB/K, we might falsely conclude that both are of concern when in fact, this tells us immediately that the walks are the only issue here in the ratio.

So, let's look at his walks per plate appearance.  Last year, he averaged a walk about 16% of the time he came to the plate.  This year, that number is cut almost in half as he's averaging a walk about 8% of the time.

What's interesting, though, is that his career rate even before this season started was just 12%.  That means that the amount he exceeded his career walk rate in 2004 is exactly the same amount he's falling short of it in 2005.  It's therefore possible that even last year, his real skill actually was around 11 or 12% and that he just had that year when everything goes right and he takes more walks because he gets more balls thrown to him, by fluke more than by pitching plans.  The nature of deviations is such that in any given year, a player will almost certainly perform above or below his real ability and almost never at exactly his real, hidden ability level.

I therefore propose that as different as these ratios look, Chavez overachieved in walks last year and is underachieving in walks this year.  With a rare stretch in April, he looks to me to be just as selective as he always has been but is getting more pitches to hit, partially because he was slumping early.  That is, rather than getting pitched around this year, I think he's getting pitched to more than he has been because he started the season so slow.  His contact rate reflects that when he's pitched to, he's putting the ball in play at the same rate as a year ago which to me indicates no real over-aggressiveness or unusual level of swinging at pitches out of the strike zone.Anyway, I've showed you that the reason this ratio has changed so much has nothing or little to do with the strikeout column.  The concern would be if the walks had gone down and there had been a corresponding decrease in contact rate, as that would indicate a newfound lack of patience on the part of the hitter.  I believe that there's nothing wrong with Chavez and that he's just spent the season recovering from a terrible start.  Since June 1st, he's hit .325 with 16 home runs in 264 at bats and to answer your final question, his future looks bright to me and though it's far too early to make forecasts for next season, given his age, I expect right now to project a better year from him in 2006 than the one he's likely to finish with this year.

From issue published August 31, 2005:

Q. Lou Pinella has publicly stated that Delmon Young will not be among the call-ups Tampa makes when rosters expand Thursday. Your latest forecast of him has him getting 16 AB. Why do you have those AB forecast against Pinella's word? I have the #1 waiver in a keeper league so I very much want to get him called up.

A. I often emphasize that my projections are what I call the "midpoint of expectations" and you're absolutely right that the Devil Rays say Young won't get the call. The problem is that if the season were played the rest of the way a billion times, my belief is that Young wouldn't average 0 games and 0 at bats. Three outfielders could break a leg in a week and he would get the surprise call out of necessity. Other circumstances can happen but basically, I have to reflect that there is a chance he could get a call. Obviously, if I believed that he was going to get called up as a September call-up, I wouldn't be projecting 5 games and 16 AB but rather 25 games and 100 at bats because an active Young would play a lot.

So, no I do not believe Young is going to get called up but I have to reflect the chance of a call-up and my projection does that. In a week or two, it will be clear that even the most severe injury to a current player would see Young not get called up and in such a case, I will remove what little there is left for his forecast. For now, I have to list something because if we played the rest of the season a billion times, with different outcomes each time but starting from today with the Devil Rays' emphasis that Young isn't getting the call, Young's average result would not be 0 games but something higher. That's why I've done that.

By the way, you talked about having the #1 waiver pick. I assume you didn't have it when Felix Hernandez was available as I mentioned a couple of weeks back that even if you wanted Young, you should take Hernandez. Also, if you're talking about 2005 performance only, the current Young forecast is not at all of any value to a fantasy team.

I wouldn't wait for him but at the same time, don't throw away your #1 pick on someone you won't keep anyway.

From issue published September 7, 2005:

Q. I drafted both of my teams this year using your 78% rule and did really well. I was leading for most of the season but the past two weeks, I've really started to slide - some might say choke - and I've fallen back to just a half point ahead in one league and back to second place in the other. Did I do something wrong? I was ahead by 20 points at one point in a ten team league just a month ago but things seem to have really derailed lately.

A.  It could be just a case of the long run having a chance to catch up with you.  Sometimes your team isn't as good as you think and other times, even a bit of bad luck at the end of the season can be the worst thing that can happen to both your team and your morale.  If you haven't committed any of what I always call "The Six Losing Sins" (you can read the essay of the same name at the main website), then all you can do is proceed with your plan.  The one most common exception that I've found, where people really do over-manage their team out of first place, is when they react too quickly to trends.  That is, when Eric Chavez was still hitting under .230 or Barry Zito's ERA stood at 4.50, did you start to think to yourself that you had a good team that only needed to be tweaked by benching the slumping players?  If so, that can often be the reason you don't have the endurance to stick it out.

A reader, who didn't want me to mention his name, actually sent me an interesting note this past week that I was prepared to put in the feedback section but I'm not going to quote him directly and his note well applies to your situation.  He said that he just fell out of first place and looked back on his season to see where there were any "holes" in his fantasy baseball approach.  He pointed out that when Andruw Jones, Hideki Matsui, Barry Zito and Vernon Wells were all really struggling, he decided that his team was good (he says he was still in 1st place in early May despite this) but that these players were drags on his performance and so he benched them at various times.  What ended up happening is that he didn't have them back in the lineup until they had already gotten hot.  He sent me a stats sheet that showed that he's missed 11 of Andruw Jones' home runs, a stretch of Vernon Wells when he hit .450, four solid starts from Zito in a row that weren't easily noticed becuase his ERA remained high at the time and most of Hideki Matsui's surge after a terrible May.  This reader told me he's in second place now but that he felt that there may be a seventh "sin" among those I mentioned and that is that he over-managed his team.  I directly responded to him and pointed out that "being too quick to react" covered what I think his mistake was here.

In other words, if you followed through with your plan and modified it only when you had enough reliable data to do so (obviously I suffered through a month of dissatisfaction emails from readers who thought my job was to project Andruw Jones' April performance), then all you can do is leave it to the performances to decide.  It is possible that you can do everything right and still not win as you're playing a percentage game.  Just don't change your approach instantly and entirely when one or two bad weeks hit.  The only way to choke is get overly worried about losing and then completely abandon everything that made you successful in the first place.

Q. I enjoyed your new relative confidence model this year that talked about the high and low range for each category. What I'm wondering is that at the beginning of the year, you said that because it was your first year of publishing such ranges, you couldn't yet say what percentage of players would fall within the range. I'm wondering if you have any sense of that now? For example, are you able to say that "60% of players will finish within their high/low range for stolen bases" for example with any degree of confidence?

As it stands now, I'm starting to get a sense of how well the ranges worked and while I can't say whether it will repeat with the same typical degree of confidence in 2006, it does seem like most categories are on a pace to have about 70% of the columns fall within their RC high and low range, which is about 5-10% less than I would have hoped.  When I set up the system, I initially used an 80% model to set up the original standard deviations, fully aware that I couldn't count on the 80% as I didn't have enough data to make a reliable estimate.  For example, as it stands through games completed last Thursday, I had counted that 29% of the players were on a pace to steal more bases than their Opening Day high rating or fewer bases than their Opening Day low rating. In home runs, about 30% of players were on a pace to finish outside of their established range.  That same 30% showed up in most columns.

So, if the 70% pace holds up, and it's pretty consistently 70% across the board right now, what it would mean is that if you looked back at the hitting forecasts, for example (and the same would apply to the pitching forecasts), of the nineteen columns you had listed for a hitter, about thirteen of them, on average, would be columns in which the player performed within the listed ranges. To me, that is far too low to be useful as the idea behind the ranges, while not intended to be actual barriers that couldn't be broken, was to attempt to give a clear picture of the lower and upper ranges in most typical situations. That's why I had been aiming for something closer to 80% rather than 70% but the only way to develop such a system is to constantly refine it based on how it does, much as we've done with our player forecasts over the years.

What is good news is that the percentages are working consistently so those with wide ranges and those with tight ranges are performing equally well, meaning we've at least achieved the beginnings of what I wanted and that is relative ways of comparing how safe a projection was so if you saw a player with 10-20 home runs as his range and another with 5-40 home runs, the ranges have contained the players equally well so that the ability to compare volatility of forecasts has at least started to function the way I had hoped.

You have to understand that some of these listed ranges were quite tight and others were completely wide, particularly for non-established players. For example, a player like Bobby Abreu was given a runs scored range of 104-120 whereas the player above him alphabetically, Brent Abernathy, was given a runs scored range of 0-52, reflecting his lack of a clear position.

What is also interesting is that in all categories, where a player falls short of his low or exceeds his high, the margin is very close, and it shows me a need to perhaps widen both ends just slightly as few players have exceeded the range in either direction by more than about 5 units. For example, if we had widened the home run ranges for all players to be +3 more on the high and -3 on the low listed range (i.e. three home runs wider on each end), only 13% of players would be performing outside the range rather than 30%. The danger there is that the wider you make it, the less you are really telling the reader something that he doesn't already know (i.e. if I tell you that Albert Pujols has a high home run range of about 80 home runs, I haven't really told you something you couldn't have guessed yourself and added value from the information becomes non-existent). The more you tighten it, the more precise you become in the ranges but the less likely it is to successfully contain a performance.

Anyway, to be honest, though I'm reasonably content that the "relative" aspect of things is working correctly, I'm still not that satisfied with how the RC Low and RC High have performed this year in our first year of publishing them and I will be looking for ways to make it far more successful for 2006.  I'm going to raise the bar from my previous 80% goal (which is unlikely to be attained in 2005) to something higher, pushing 90 or 95% if possible. In particular, I want to tighten the ranges for non-established players and widen them slightly for veterans to make them both more useful and more capable of containing the actual performances. It was our first year of doing that aspect of the forecasts and as I said when we ventured into that territory, when it comes to confidence ranges, they are very much to be treated like our first forecast sets published back in the early 1990s as these sorts of ranges are in their infancy and thus, we haven't built up much of a track record. I've set aside some time this off-season to research a better, clearer and more useful confidence model as I really wasn't satisfied with this first effort and it certainly didn't add as much meaning to the forecasts as I hoped it would. Readers know that I'll say when I'm not happy with my own work and on confidence range research, I don't think I've really achieved much in 2005 even though we have a starting point and a slight increase on both ends would clearly and easily improve the ranges significantly.

My work and research on confidence ranges continues.