The Best of Ask Baseball Notebook: Volume 4
published Wednesday, November 15, 2006

In this fourth "best of" collection, we offer selected questions and answers from the Ask Baseball Notebook series, published in the past year or so.  Questions are not necessarily selected for how good or bad the quality of the advice turned out to be but rather how well they illuminated a common concept or topic that came up.  So, in some cases, selected advice may very well have turned out to be wrong but the published response revealed something about how our forecasts are made or the way we approach the process of valuation, for example.  Unless otherwise noted, David Luciani was the responder to listed questions.  Selections are in chronological order by publishing date.  Newly-added follow-up commentary for selected questions is in red bold print here.

From issue published November 17, 2005:

Q.  I picked up Roy Halladay in a trade while he was on the DL (never appeared for me and I still won).  How do you think he'll bounce back from his injury?

A.  One of the things I really like about Halladay is that he has a tremendous work ethic.  Moreover, while he may seem injury-prone at first glance, you discover that the 2005 injury was a fluke where Kevin Mench hit a ball to break Halladay's shin.  That's not the sort of nagging or recurring injury that can threaten future seasons so while he hasn't started more than twenty-one games since 2003, he was a workhorse for those two years of 2002-2003 and he saw how quickly his team faded from a remote playoff hope to a distant third place team once he got hurt this past season.

I think I mentioned in this space last year the work that Sig Mejdal had published in a previous edition of the annually published Bill James Handbook and how I was remotely envious of the quality of work he did on injuries in that I thought he had begun to form a more accurate model than mine on forecasting injuries.  In fact, he inspired several revisions to my own model for 2005 that will also apply to 2006.  In short, and at the risk of oversimplifying the quality of his research, Mejdal concluded that for both hitters and pitchers, "past injuries... are a great predictor of future injuries" and while that may, on the surface, seem obvious, Mejdal broke it down into three primary types of injuries that serve as predictors for pitchers.  They were elbow injuries, shoulder injuries and arm injuries.  In the case of Halladay, this 2005 injury was a fluke ball that broke a bone in Halladay's leg.  Inspired by Mejdal's research, I tried to find any correlation between broken legs in one season and recurring leg problems in future seasons and I couldn't find any reasonable connection that was useful to forecasting.  I expected there would be one, perhaps out of a bias of my own as I broke my own leg in two places when I was a child and years later can occasionally feel stiffness in that leg but the scientific proof of it as it applies to baseball simply didn't exist in any of the data.  I found mild relationships between pitchers who had suffered from recurring leg soreness that was strong enough to land them on the DL but I couldn't connect the actual fluke breaking of a leg to future leg injuries enough that they would serve as useful predictors.

Now, Halladay does have the more significant shoulder history in that he missed much of 2004 with a tired shoulder and that is the kind of problem that would be more likely to recur.  Regardless, the weight of his injury history (tired shoulder in 2004 preceded by two years of 34 and 36 starts in 2002 and 2003 respectively) isn't enough to cause me to forecast him to miss much time in 2005.

Putting aside the health issue, then, Halladay's stuff has really never been better.  If he hadn't have been hurt the week before the All Star game, Terry Francona admitted that he would have started the All Star game and though we can't know for sure, he was well en route to another Cy Young Award.  His ERA for the past five years combined is an amazing 3.16 and his control continues to improve.  Best of all, he doesn't even turn twenty-nine until May.

I'd say the chances of Halladay having a strong season in 2006 are excellent and these factors, mixed with the expected improvement of the Blue Jays, who now have an extremely-expanded free agent budget and should be a better run-scoring team, make Halladay an excellent bet for 2006.  You were smart to pick him up when you did.

Halladay ended up having his healthiest year since 2003, starting 32 games and going 16-5 with a 3.19 ERA before being shut down early in September.

Q.  What is the best way to utilize your service pre-draft, draft day and throughout the season?  I am in an NL only 5x5 league.  Do I, for example, use the projections in conjunction with previous standings to decide how much of each category (runs, strikeouts, SBs, etc.) I need to acquire?

A.  This is a question I really wish someone had asked earlier the way you have here and I've been tempted to write an unsolicited essay on it before.  My response here, therefore, will be the primary focus of this issue because I really think readers will benefit from it.  I'll try to offer references to various essays and answers I've already published on the details of my answer as I think that will help you more if you're not comfortable with a particular concept.

I'd love to tell you that there's one method to use the service but everyone's league is different and more importantly, everyone has a different style about how they use information.  Some people like to consult multiple sources and others like to first form their own opinion and then check another source to validate or second it.

All are good ways of using our materials and a lot depends on the type of service you subscribe to as we have several packages.  I'll presume for hypothetical purposes that you subscribe to all the features of our site (including the in-season minor league translations) and I'll tell you how I use the site myself and I'll try to take you from the point when the first projections are published up until the point of the final projection set.

Last year, I published my first player forecast and ranking sheets in early December and the moment they go online, I start using them for my own keeper leagues and, in the rare occasion of a single season league that drafts early, I start to think about strategy.  I put in my league parameters into the ranking sheets and create my Internet Explorer bookmarks to the result so I only have to do this once a year.  If you haven't noticed, if you bookmark the results/ranking page, you don't ever have to re-enter the parameters and the ranking sheets will automatically update with new values based on each projection set.  Another thing you might not have noticed is that below the regular rankings, you get the rankings by position based on your league qualification rules.

So, this first ranking sheet gives me a preliminary list of player values based on that league.  One of the first exercises I do every year is to compare those projected values to the salaries I will have to pay for the players I may keep.  I explained this method in exact detail in the column called Keepers and Optimal Bidding, which you can read for free at the site.  Basically, I'm using my model that says that in a typical league, you need to acquire players at an average of about 78% of their projected value in order to win a league.  The 78% number was not arbitrary but rather came about from analyzing thousands of leagues (originally it was hundreds and as we increased the sample to thousands of leagues, the 78% number held up).  If you'd like more about the concept, you can read that in the essay called On Paying 80% of Projected Value, also available at the site.

At the risk of repeating what I explained in the keeper essay, I'm trying to assemble a keeper list that takes me into the draft with a good chance to pay 78% of my value for my team.  It does not mean that I'm going to limit myself to protecting only players that are earning 78% of their projected value.  Occasionally, you can get a player who is so underpaid compared to his projected value that he literally pays for me to keep a superstar who is actually overpriced.  A $5 Bobby Jenks, such as the reader in an earlier question in this space said he has, may be enough if he ended up as a projected full-season closer to enable me to protect a $35 Mariano Rivera projected to be worth only $30, hypothetically speaking.

So, with those first projection sets, I'm looking to see if there are players who, either because of their salary or projection, are players who are not going to fit into my keeper list.  This is crucial because there may be many players on this list who other teams will value as potential keepers or for leagues that use it, for so-called topper rights.  Sometimes, if I project a collapse compared to the previous season, I may have a player who is valued so much more than the projections do that I have to trade him.  What's great is that if I've done the exercise I talked about in the keeper essay, I can target players on an opposing team who may be undervalued by the opposing owner.  This is an effort to constantly expand the amount of value I can protect and therefore, be able to go into the eventual draft with a bid modifier that is better than 78%.  A 78% bid modifier makes it extremely tough to pick up any superstars and while it's not necessary in order to win a league, a good keeper list can sometimes send you into the draft able to bid 90% or even 100% of the projected value because your keeper list is built so well.

Once keeper season is over, I then create a new draft ranking sheet.  I set the bid modifier in the online ranking tools to match the one I need to use for the draft based on the quality of my list.  As a subscriber, you'll have noticed that our ranking sheets let you do that so, let's say that I set it at 85% because of a pretty good keeper list that affords me a bit better than the low 78% target.  The ranking sheet prints me the list with both the projected value and the recommended bid based on the 78%.  It doesn't mean I'm going to bid 78% for every player and if you enable the "forced escalation" option, which I strongly recommend, it will often adjust the bids at the higher levels so that the top players get a slightly better projected value and there are minor adjustments made for scarcity of statistics - not to be confused with position scarcity - where forced escalation is a recommended option even for draft leagues.  I personally use it for all of my leagues now because it has worked so well and in 2005, I'll confess a secret that will make it tougher to win next year (as I know at least a few of my opponents read my materials) but it was the first time I've ever tested forced escalation in a straight draft league as opposed to an auction league and it helped me win the 20-team Mixed Nuts Expert league by 20.5 points.  Even better, the number two team was also a Baseball Notebook writer in Seth Trachtman so I felt pretty good about this outcome.  That's the league where I "half-accidentally" punted saves in the draft (though I picked up several closers later in the season) because the draft sheets and the forced escalation adjusted for the scarcity of other categories and virtually recommended this approach to me for this league.

So, I have my draft sheet printed and what I do is I also become very comfortable with the previous season actual performances and the general perceptions about players.  I tend to be already because of the frequency with which I see statistics but if you're not comfortable with last year's performances, at least take the league leader lists from last year and become comfortable with who had an exceptional year.  Just as I explained in a question about Greg Maddux before 2005 (which is included in the recently-published of "Best Of Ask DL: Volume 3" at the site), I'm trying to have a good sense going into the draft/auction about what my opponents will think about players because I never want to take a player earlier than I have to, even if it means I miss out on someone if I underestimate what my opponents will think about that player.

Now, let's say we're up to draft day.  I explained in the essay called Being Prepared for the Fantasy Draft how to go into draft day and all of the points I made there still stand.  Let's say the bell has rung on draft day.  Without falling into the so-called "alphabetizing trap" I mentioned in that preparation essay, here's how I target certain players.  If it's a straight draft, as opposed to an auction, I'm looking for the top-ranked player whose perception is such that everyone else will think he's valuable too and thus, he won't last until the next round.  So, looking back at my actual Mixed Nuts League draft sheet from 2005, I had the following listed as the top five in order: Pedro Martinez, Albert Pujols, Alex Rodriguez, Randy Johnson and Alfonso Soriano.  Now, I was picking 10th in a 20 team league and while I wanted Pedro, I knew that no one else would value him as highly as I did and his projected value for 2005 was much higher than his 2004 value so I bypassed him for the first round pick.  Pujols and A-Rod were both gone before my pick came up and that left Johnson, so I took him.  We all know he didn't have as a good year as some expected (including me) but following this strategy, with my second pick - this was a snake draft so I was picking 11th in round two - I decided it was a good time to take Pedro, my top pick, as I felt he would probably go before my next pick in round three.  Using this approach where I attempt to estimate how late I can afford to take my top remaining pick, I always do miss out on a highly-ranked player or two because I make a mistake about how long they will last.  I had Andruw Jones listed as one of the top players after this top group and I was prepared to take him in round five of that Mixed Nuts League when to my great dismay and amazement, Paul Sporer (who writes for both Sportsblurb.com and Rotojunkie.com) took Jones with the pick just before I was to pick in that round and I had to settle with John Smoltz as my round five pick instead.  It didn't bother me because I'm playing the percentages and that is that I would rather miss a highly-ranked, highly-projected player, even one projected for a breakout year like Jones was, than take a player too early and waste such a pick on someone I could have picked later.  For every Jones I miss out on, I'll also miss out on a Greg Maddux who disappoints.

So, in draft leagues, you're trying to take the player who is highest-ranked on your list but only if you believe you're taking him in the final round in which he would be available to you.  You need to be confident, if not certain, that it's the final round and if you're in doubt, go ahead and risk leaving him out there.  I had no doubt that Randy Johnson would be gone by round two when I took him and I had little doubt that Smoltz was high on everyone's list when I settled on him the moment Andruw Jones disappeared from the pool.

If you're in a draft league that requires submitting your list in advance, it's a bit tricky but here's how I make my list.  I tend to view the ranking sheets from the site in terms of blocks.  So, let's say my top ten for 2005 had said this in terms of the actual projected value: Pedro Martinez, Albert Pujols, Alex Rodriguez, Randy Johnson, Alfonso Soriano, Andruw Jones, Derrek Lee, Bobby Abreu, Manny Ramirez, Roy Oswalt.  Viewing this as a block, I want to move the highly-perceived players higher on my list and then get those others listed where I can still get them.  Thus, a list like this I might have re-arranged to look like this: Pujols, A-Rod, Johnson, Soriano, Abreu, Ramirez, Oswalt, Pedro, A. Jones and Lee.  I actually recommend wider blocks than a block of ten depending on your league and a block of fifty or even one hundred isn't unreasonable if you're in a ten team Scoresheet league, for example.  Basically, you're getting those high-valued, high-perception players near the top of your list and those high-valued, lower perception players behind them.  If you end up with a lower-ranked, higher perception player, you have little trouble moving them in a trade for the player you actually wanted and you can probably get more in the deal.  So, if you had Bobby Abreu ranked eighth and got him and missed out on a sixth-ranked Andruw Jones, you wouldn't have had too much trouble trading Abreu for Jones right after the draft if you've adequately estimated perception.  Of course, if you think Jones was more highly valued than Abreu, then he should have been listed in the proper order on your list to account for this.

Let's talk about auction leagues.  There are several things I'm watching for during a league auction.  First, I don't want to accidentally fill a roster spot that I had hoped to fill with someone stronger.  So, if an opposing fantasy GM brings up a player who is, in my mind, mediocre or far below the quality of talent available, I'm not even going to join the bidding for him.  I suppose it's my own form of position scarcity consideration.  It's not that I'm modifying the projected value of the player but I am allowing for whether a player is going to plug a spot that I could otherwise still fill with much better talent.  There are rare exceptions and sometimes I don't mind if I pick up a $10 player for $2 but in general, I need to have a good sense of my target players and basically, I'm only going to bid on a player if he wouldn't fill my roster (as in I have only one outfielder, five outfielder spots to fill) or if I think he has some moderate or good projected value.  In the essay in which I explained the 80% value approach, I explained how to adjust the bid modifier during the auction and it's my own response to those who think that you have to account for draft inflation by overpaying for players.  The problem with the commonly-understood concept of draft inflation is at the end of the draft, the players you bought then instantly revert in perception and reality to their projected actual value and not the inflated value you created during the auction.  However, if you use the adjustment method I talked about in the essay, you'll see that you create your own ability to inflate bids because of the bargains you pick up in the middle parts of the auction.  Use those to your advantage in conjunction with the sheets and you will be able to outbid everyone and overbid in the style of someone who considers draft inflation but in your case, you'll be doing it to assemble real, ultimate value and not simply perceived value that comes about as the result of most fantasy GMs propensity for overspending early in an auction.

Okay, so by the end of the auction, I've targeted having my 78% value-paid roster.  Contrary to expectations, I'm not spending a great deal of time at this point thinking about whether I've overbuilt one category and am weak in another.  In most leagues, I tend to let the standings play out until at least a third of the way through the season before I see if there's a category where I'm too strong, and by too strong, I mean that I'm wasting the accumulation of a category to win it by too wide a margin.  With my competitive spirit, I want to finish first in every category but I don't want to win any category by more than I need to even though sometimes this will happen by the surprising good fortune and performance of players.  I am not one to put great emphasis on adding the year-to-date standings to the projected values because if I'm leading a category by a wide margin, that motivates me to trade it, often where I will then remove that category from my league rankings to see how the players should be valued if I disregarded consideration of that category.  I did this in several leagues last year when I was leading either/both of the wins or strikeouts category by a wide margin because I had more quality starting pitchers than anyone else.

So, from the time the auction ends until I run into situations where I'm dominating a category by 30-50% of the way through the season, my simple focus is on acquiring players whose projected remaining value is higher than players who I have on my roster.  This is not a simple mathematical exercise because I am not just going to pick up slumping players and trade hot ones because their projection is strong.  Rather, I'm going to demand as much in a trade as my opponent would think I would want based upon the general perception of those players.  That is, the trade needs to not just benefit me in terms of the projected remaining values but it needs to be close (but obviously not at par if I'm to actually get my opponent to agree to the deal) to appearing like an equal trade in the perception of those people who tend to overvalue April and May performance.  For example, I picked up Andruw Jones, Barry Zito, Eric Chavez, Vernon Wells, Randy Johnson and Hideki Matsui last year all in separate trades and leagues last year but I demanded more from my opponents than I normally would and more than the projections would have said I needed because the opponents, though they might not admit it, were actually afraid that the players really were not as good as they thought and really did want to trade them.  In one league on the final day of May, I was even able to acquire Chavez for Shea Hillenbrand straight up.  I would have been willing to trade more than Hillenbrand but I didn't need to because Chavez's perceived value fell so low that I was able to account for it in combination with what I thought was excellent projected value for the remainder of the season.  Chavez went on to hit .293 with 23 home runs and 78 RBI the rest of the way while Hillenbrand hit .276 with 12 home runs and 56 RBI the rest of the way.  In hindsight, the deal seems obvious in my favor but you have to remember that Hillenbrand was far off to a better season than Chavez at the time and using the ranking sheets projected value wasn't enough to make the deal.  It was that I felt I was getting enough perceived value in return in that I felt my opponent would think that Chavez was barely less valuable than Hillenbrand.  Because my opponents know about my column, publishing this here always jeopardizes my future ability to do this but in short, I am always in search of the higher projected value player who is just that - barely valued less than the player I am more than willing to give up to that opponent.  I've underscored the incredible lack of accurate perception that happens with slumping players early in the season in the essay called On Fantasy Baseball Trading which I strongly encourage you to read if you haven't already.

Another clarification is required because a few readers misunderstood what projected remaining value means.  Projected remaining value is relative based upon the league base amount, which is essentially the total value of dollars left to be spent in your pool.  Once the season starts, it serves as a base for calculating relative value but the ranking sheets can't possibly know when your season started - some people play in half-season leagues for example.  So, if you see a player projected to have $40 remaining value on August 1st, it's obvious that this is not comparable to the $40 you would have paid on draft day if your league started in April.  I had several readers write to tell me that they couldn't understand how a player could be projected to be worth $40 for the month of September and my response is that the value listed is relative.  Years ago, we used to shrink the total base as the season went on but with the proliferation of mid-season leagues, it didn't make sense.  Moreover, if we do shrink it, the values become unreadable by mid-August because then the difference between a top player can be just $1 and this grouping becomes impossible to read, even when we show our non-rounded dollar value projections.  It should not affect your decisions because the value you're looking at no matter what point you access the sheets is and should be how much value a player has to your team the rest of the way, relative to other players available and the sheets do calculate these as they should and as I want them to if you're to have the best chance at success.  Of course, if you want to force the sheets to show you actual remaining value that's adjusted for the portion of your season already gone, you could simply lower the base by lowering the salary.  You'll see how pointless this is if you try it because it doesn't benefit you - once the season is underway, those salary limitations no longer apply and value remains relative and in the same order.  Don't get distracted by such concerns and leave the base as it was to see the variation in values the most clearly.

One other element of success in using the sheets is to search the waiver wire for players who have higher projected remaining value than players I have on my roster.  For this exercise, perception is meaningless unless you have to make an educated free agent bid for them.  In general, you simply want to get those players that are out there for free or cheap who have higher projected value than a player you've been keeping.  I've won many leagues based on my waiver claims alone and key pickups last year that you should have been able to get early based on the projections were players like Todd Jones, Robinson Cano, Derrick Turnbow, Fernando Rodney and others.  It's the one time when you really don't have to do much thinking because the projected value will tell you a lot, unless it's so close between two players that you must make a decision based on your experience and intangible opinion.  I have occasionally traded a player I was about to dump to make room for a free agent claim and that's usually a good way to free up a spot if you think the player who you would drop could net you another useful part in a trade.  Of course, in most leagues, you can't time it that way so what you need to do is you trade your active player, temporarily bring up a player from your reserve list who can occupy the vacated spot, make your claim for the player you really want and then when the player is awarded to you, send your reserve type back down.  It's risky but if you time it right or judge the free agent perception market accurately, you can pull off some excellent deals without wasting the dropping of players who can actually help you via trade in another area.

So, let's say now that we're at mid-season or even late.  By around June 20th or 30th, I'm looking to decide if I'm in the race.  If it's a single season league, then no decision is required.  I simply have to keep playing for this year.  If it's a keeper league, I need to make some sort of informed decision because if I'm not going to win, then I very well better win next season by planting the foundation right now.  This is where the minor league translations come in.  I tend not to use them too much when it comes to players getting called up because the projected big league values already account for their current projected ability.  Granted, there are exceptions because a player might get called up on a Wednesday and I can't wait until the next projection is finalized on Sunday morning.  As the brain behind the projection, I can "cheat" in that I can make the projection earlier but since readers can't do that, they may then have to rely on a glance at the player's minor league translation to see what sort of season the player is having and whether the translation implies an immediate big league readiness.  I picked up Dontrelle Willis one season for this very reason because a few years ago, he had that rare combination of high translated ability and an organization that was close to giving him a chance.  If I have them ready, I also tend to refer to my BNRA projected callup chances at this point because they have served me so well in identifying who is close to getting the call.  I tend not to use these numbers too much if I'm a contender because I'm worried about winning this year and I've already accounted for these percentages when I've made the actual big league forecast for the remainder of this season.

When it comes to keepers, though, if I decide that I'm out of the race, then I'm going to be looking at the translated minor league leaders to see if there are players near or at the top of the lists who are possible September call-ups.  It's not that I care much about September if I'm out of the race but the importance of September is that it is the most important period in terms of winning a full-year job for next season and rookies who don't get the call in September (such as Delmon Young this year) face a much steeper climb in the subsequent season to win a full-time job.  It's not that they're valued less by their organization but rather, the weight of spring training and the short sample it offers makes the variation of performance so wide that even the most highly-touted prospect who didn't get a call the previous season will lose his apparent full-time job if he hits .150 in the spring.  That's not true of the prospect who hits .150 in the spring but hit .350 over twice as many at bats during the previous September.  So, if I'm in that position, I'm targeting acquiring such players who are likely to get a September call-up and whom I can pick up for the minimum salary, either through trade or on waivers.

I have to admit that by around September 1st, the work is essentially done if I'm still in the race.  A team out of the race will still be scanning for players to add for future seasons but typically, if you're high in the standings, you're not going to want to risk too much on players who might not be ready and who are getting their first look.  The rare exceptions, like Felix Hernandez, will have been taken by another team who had a higher pick than you because they're not doing as well so you're not going to get them anyway.  Basically, the key to the stretch run is no longer the forecast but maintaining your activity to ensure you're maximizing plate appearances and for pitchers, starts for starters and appearances for relievers.  This includes consideration of the weekly schedules, such as we cover in The Scheduled Advantage series that is published every Sunday during the season, and it also means that you never commit any of what I call The Six Losing Sins, which really are murder to any fantasy team that was otherwise destined to win its league by a tiny margin.

Finally, some readers have asked me whether I will mortgage my future to win a league this year.  Perhaps I'm arrogant but I always believe I can win future years even when I do trade away highly-valued prospects.  In fact, if you want to do a fun exercise, look back at the top ten prospects published by most publications in previous years.  While we all like to think we're good at it, there are so many so-called "can't miss" prospects who never pan out or, worse to a fantasy team, take about three full years in the big leagues before they really learn how to play up to their potential.  I tend to believe that if I'm a contender, and I always think I am until both the standings and the calendar tell me otherwise, I'm trading almost exclusively for this year, confident in my ability to rebuild a depleted future during the off-season and focused on the single and repeating task of winning the current challenge.

Above all, don't follow anything so strictly that you ignore your own opinions and feelings.  Remember, your instincts likely come about as a culmination of everything you've read and learned about baseball.  If something feels wrong to you and you do it in blatant defiance of your instinct, there's probably something in your mind that knows a factor that perhaps the published forecasts or player values can't consider.  Maybe your league has an element in it that no player value can learn.  Maybe you only half-recall a conversation you had with an opponent about a player you're thinking of trading.  By all means, use the information available but always use it to supplement rather than replace your knowledge.  You'll find that while you might miss out on deals that would have benefited you, you'll get the most enjoyment and the greatest success if you combine collected information with what you already know and have learned about your unique situation and league and your own understanding of baseball.

I hope this answers your question in a way that you find both useful and applicable and even for the readers who don't subscribe to some of our online tools, I'm confident that the free materials I've referenced and made available to you here will help you in your annual fantasy challenges.

This lengthy response to a single question could easily have been entitled as its own essay on how David Luciani manages his fantasy team.

From issue published December 20, 2005:

Q.  I found your first projection for Ryan Howard quite interesting.  I found the fact that he hit only .148 in 61 at bats vs. LHP and his 33/100 BB/K ratio disturbing.  Do you see him making similar progress to David Wright with regard to plate discipline?  What factors led to his strong projection?

A.  It seems that you apply more weight to some factors than I do.  For example, BB/K ratio can be crucial when it comes to forecasting playing time but the trade of Jim Thome and new Phillies' GM Pat Gillick's open commitment to Howard as the everyday first baseman make that only a minor concern.  It's not often that the reigning Rookie of the Year loses his job within one season and so, I don't think the playing time aspect of the forecast is in any way risky (149 games, 536 AB in the first set).

In fact, I talked before about my disagreement with applying too much weight to ratios like BB/K ratio.  There is no doubt a strong correlation between ratios like BB/K and, say, hitting ability or run production, but my view is that it's because the denominator of the equation, strikeouts, is key to the contact rate and thus the batting average of a player.  Howard does strike out plenty and in fact, his strikeout rate is why he hasn't often made it to the highest level on my top prospect lists.  That is, I can see him having a decent but shorter career as a good home run hitter but one who won't have the longevity into his mid thirties because his propensity for striking out will make him fade away quicker than someone like a Manny Ramirez, who still has typically struck out more than 100 times a season but not to the degree that Howard does.

Just looking at that first projection for Howard, I projected a .280 average with 38 home runs, 63 walks and 202 strikeouts in the first edition of the forecasts.  The strikeouts would be record-setting but it is entirely possible to hit .280 while striking out that many times, if your bat speed is good enough that when you do make contact, you hit the ball sharply and with authority.  Also, you'll notice I am projecting an even worse BB/K ratio for Howard in 2006 when you break down the numbers (0.31 BB/K compared to 0.33 BB/K last year).

Interestingly, another reader wrote me a question that was similar to yours in that it mentioned the BB/K ratio but made the argument that it is impossible for a player with a BB/K ratio below 0.40 to hit .280, which couldn't be farther from the truth.  In fact, just looking at 2005 numbers, Carl Crawford hit .301 with a BB/K ratio of 0.32, Robinson Cano hit .297 with a ratio of just 0.24, Mark Grudzielanek hit .294 with a ratio of 0.32 and Shea Hillenbrand hit .291 with a ratio of 0.33.  Those are just a few that came to mind when I was looking for examples and in fact, we found 44 players in 2005 alone who had at least 400 at bats and hit better than .280 with a BB/K ratio of worse than 0.40.

I was surprised at the number of people who asked about or mentioned Ryan Howard in questions submitted to this column recently as I actually didn't think I was out on a limb that much.  In 351 major league at bats, he has 24 home runs.  He hit 38 in just 522 at bats last year when you combine his big league numbers with his Triple-A Scranton-WB numbers.  In 2004, he hit 48 home runs between Double-A, Triple-A and the majors combined in fewer at bats than I'm projecting for 2006.  If he lost even 20% of his home run ability in the transition from the minors to the majors, he still comes out as a pretty safe 35-40 home run type.

In terms of his performance against lefties in 2005, he hit just .148 with 1 home run in 61 at bats against southpaws.  It's not too much of a concern for me though I did account for him facing a greater ratio of lefties in 2006 than he did in 2005 as he faced righties about 80% of the time in 2005 and that's more likely to be about 70-75% in 2006.  In other words, his being a full-timer now is only going to mean about 85 more at bats against lefties than he had last year (or about 145 of his projected 2006 at bats).  I've accounted for that and he would have had a much better home run projection if he could face right-handers as frequently as he did last year over the new projected number of at bats.

To our amazement, our high forecast for Ryan Howard actually sorely underestimated the type of year he would have as he would go on to hit .313 with 58 home runs and 149 RBI.

Q.  Can I ask why you have such a low at bats total projected for Delmon Young?  I was under the impression that he was going to start in 2006.

A.  In the projection notes/introduction, I think I explained the importance of representing all the possibilities in a projection and I gave an example that if a player has a 10% chance of being a full-timer playing 150 games and has a 90% chance of not even making the team, then I should reasonably give him a projection of about 15 games, representing the average outcome for the results (i.e. for the statistically-inclined, 10% * 150 + 90% *  0 = an average result of 15 games).

In the case of Young, you see this effect at work as I've forecasted 46 games and 181 at bats, productive at bats at that in that over that span, he would look like a 35 home run / 15-20 steal type with a questionable initial batting average in the majors. I'm trying to represent both the possibilities here and account for the current situation in Tampa Bay.  When we went to press with the first projection set, the Devil Rays had Carl Crawford, Rocco Baldelli and Jonny Gomes all under contract.  All reports I've received say that Baldelli is going to be ready for spring training and there you have your Devil Rays' outfield.  That means that if Young is to break in, either one of those three would have to go or the Devil Rays would have to use the DH position to play four outfielders regularly, which is possible.

Young is not guaranteed a starting spot or even a roster spot and in fact, it's going to be interesting to see how the front office responds to the open criticism Young made of the organization back in September.  Young was quoted in the St. Petersburg times as implying the Devil Rays were a second-class organization for not recalling him last September and he said "..as soon as I get my time in up there, I'll bounce out of there.  There's no reason to stay around for the long haul.  Get your six years and leave."

While the ownership situation in Tampa Bay has been in transition, I still don't think this sort of commentary is likely to earn him a quicker opportunity and I just can't see where to give him the at bats here.  Also, if I'm right that his initial ability will be a .240-type hitter in 2006 (though he remains my #2 prospect overall), that also means that even a plan to have him play regularly could go drastically wrong in spring training.  If he really is a .240 hitter right now - and I know he's going to improve quickly over the next few years - the margin of error is quite wide for him to have a terrible spring training and lose a spot on the Opening Day roster even if the team wants to give it to him.

Until things change either in the outfield situation or in terms of the open commitment the team has to Young's 2006 plans, I feel comfortable with the current projection.

Young did finally get called up to the majors after getting himself suspended for much of the minor league season.  He finished with 30 games and 126 big league at bats in 2006, not too far off from the amount of playing time we had given him at this point.  The point to take away here is that the so-called "sure things" of November and December often don't get their chance when the bell eventually rings in April and even the best rookies are high risk in terms of projected playing time if you're after a full-time player.

Q.  To what extent will Shea Hillenbrand's comments affect your forecast for him?  How can a player bring his game to the "next level" as he calls it?

A.  I rarely put much emphasis on player comments because players often think they can do things that they can't.  A good example made it into last year's "Best of Ask David" column - check out the question about Adam Dunn saying that he thought he could steal 25-30 bases for how I would respond to Hillenbrand's notions of his new capabilities.  For readers who are wondering what Hillenbrand said, he was recently quoted as saying that he had lost 18 pounds and that he had changed his swing and though it would take his game "to the next level" and went on to predict that he "can hit 30 home runs and drive in 100."  I root for Shea Hillenbrand and it will be interesting to see the new swing but in terms of the weight loss, I can note that actually, when a player loses weight, his power typically drops rather than goes up.  I actually did an analysis in these pages on that topic which you can read at the website called "Size as a Statistical Field" which showed that the average 211 pound player (Hillenbrand's approximate 2005 weight) averaged about 18 home runs per 550 plate appearances.  Interestingly, Hillenbrand hit exactly 18 home runs last year in more than 600 plate appearances.  The analysis I talk about there shows that a drop of 18 pounds to, say, 193 pounds, would actually cost Hillenbrand about 5 home runs in power, on average.  Anyway, don't put too much emphasis on his comments until you see him playing, even in spring training.

Hillenbrand essentially performed at the same level as he always had prior to 2006, if not a bit below his career norms.  His batting average dropped significantly after a trade to the Giants in mid-season.

Q.  It seems you still believe in Kaz Matsui's ability.  He's projected for 291 at bats.  Is this because of the injury risk or because he won't be the Mets' starting second baseman?

A.  No, I've dropped quite a bit in my opinion of Matsui's opinion and his projected performance is comparable to his career numbers in the majors so far and essentially don't even resemble his star performance from his Japanese days.  As for the playing time, when we went to press with the first edition of the forecasts, the Mets didn't have a second baseman on their roster as Miguel Cairo was a free agent and Chris Woodward didn't look like more than a utility player.  The only player who seemed to be a possible fit other than Matsui was Jeff Keppinger, who actually got more projected at bats in the first set than Matsui did and may well end up as the second baseman.  Yes, there is also a moderate injury risk on Matsui's part as he did miss a lot of time in 2005 and that played a role in the forecasted totals.

Matsui ended up with 243 at bats for the year, splitting the season between the Mets and Rockies.

Q.  Do you still project such high offensive numbers for Miguel Cabrera?  It looks like he will be surrounded by mostly Double-A and Triple-A players.  Why would any pitcher throw him a strike?

A.  Barry Bonds is the best example of a player whom pitchers didn't want to face and even with his record-setting number of intentional walks, pitchers still think they can get him out, especially with the bases empty.  Pitchers will throw Cabrera strikes because when they don't, he's a good enough player that he will often lay off and the pitcher will get behind in the count, won't want to walk him and then will come in with something hittable.

I do expect Cabrera to get pitched around quite a bit and that's reflected in his projection, which has him down for a career-high 79 walks, some of which is the natural improvement he should gain through development and experience and some of which is the effect of the lineup.  I still think he'll have plenty of opportunities to produce and even with an apparently inferior lineup around him, I think he's going to have a better year in 2006 than he's had so far in his brief career.

It's open to debate whether Cabrera had a career year (with the side against calling 2006 a career year probably the winning side of this argument) as he did set a record-high in batting average (.339) and as walks (86) but his home run output dropped from 33 home runs in 2005 to 26 in 2006.

From issue published February 15, 2006:

Q.  Do you ever think that sometimes you are influenced by a contrarian's view - that sometimes you might rate some obscure players a bit higher than they should be simply because you like to tweak readers' expectations?  I ask this because I'm likely going to draft two players who you seem to rank higher than anybody else does - Pence and Denker.

A.  No, I never rate players at a specific point to generate controversy.  I know my choices are often controversial but it's because I also never rate players based on what others say as I have my own forecasting method.  In fairness, sometimes it works and sometimes it doesn't but the reason I might be considered a contrarian is because, as I've often said, to me a top prospect isn't a top prospect simply because everyone says he is.  Certainly, if a player is highly rated by a prominent publication or writer, naturally that player is going to be someone I am likely to notice or even consider in much more detail than I could if I were to analyze 4,000 minor league players with the same method but ultimately, a player's position on my prospect list is still going to come down to where I think he belongs as I'm trying to be right.

It occurred to me in recent weeks that if I had always said what everyone else said or expected me to say, I could make my prospect lists much more quickly and with a lot less effort and certainly with a fraction of the criticism I've received over the years.  They would be nice, neat looking prospect lists, not much different from every other, would probably have a certain degree of reliability because consensus picks often do work out and they'd offer nothing new to the discussion.  I'd go away feeling like a hypocrite but I'd deliver what's expected rather easily and without getting the usual angry response prospect lists seem to cause.

I tell you what I think and if it's contrary to popular opinion, at least you know you're getting another voice in the discussion who tells you what he really thinks.  I've learned, particularly recently, that a lot of people don't think much of such an approach but I'd rather tell you what I think than tell you, in summary form, what everyone else thinks.

As for the two you mentioned, I know I rank Hunter Pence higher than most but he belongs there as the model tells me he projects as a future .295 type with 30-40 home run power.  Denker projects as an eventual .280s type with 30-35 home run power and in his case, an excellent future on base ability that should keep him around for a long time.  I think both belong near the top of everyone's prospect lists, where they appear on mine.

Hunter Pence will now move to the top of most pre-2007 prospect lists as he had a breakthrough year in the minors, hitting .283 with 28 home runs and 95 RBI in 136 games, setting himself up with a possible chance to crack the Opening Day big league roster in 2007.  Travis Denker did not, however, as he hit just .247 with 16 home runs and 70 RBI in 441 minor league games this year.  The point of including this question here isn't so much to talk about the two prospects as it was to respond to whether our so-called "contrarian" views are deliberately designed to go against the grain.

Q.  I've noticed watching my NL-only league for the past five seasons that the split is about 70-30 for prices actually paid for hitters-pitchers.  As a result, I end up with a ton of bargain pitchers every year and weak hitting.  Should I use 70% for hitting when making my ranking sheets?

A.  Yes, I recommend you shift to a 70/30 model if you're positive this is the case as you want to allocate your split approximately the way your league does.  A lot of readers write to tell me that they sincerely believe that a 50/50 approach is the right way to go, despite how people spend their money, because they argue that if there are five pitching categories and five hitting categories, that you should dedicate half of your resources to each.  The problem with this is that in fantasy baseball, there really are essentially two different team budgets, one for hitting and one for pitching.  I've experimented with the 50/50 approach many times and in many formats against both extremely strong competition and against the weakest of fantasy players and it usually doesn't work.  You end up with a strong pitching staff, sometimes even a first place pitching staff, but a mediocre to even below average offense and you end up finishing 6th or 7th overall and you wonder why.

I'm very confident that you should weigh the categories the way your league will spend its dollars between the two and the best predictor of that, especially if you have the same people returning year after year, is to look back in recent years to see how the resources have actually been spent.  Our fantasy domination sheets have a parameter for this so simply change the number 65 to be 70 will be all you need to do and you'll get appropriately modified values.

This also offers me an opportunity to address an older question someone asked about a draft league, where they said that there weren't enough hitters highly ranked on the sheets in the context of how the player's league values hitters.  Even if you're playing in a draft league and not an auction league, you may still get some sense, overall, of how your opponents value hitting over pitching.  Lacking the dollar values an auction league offers, I often will tweak the number anyway to attempt to reflect how a league balances hitting and pitching and you might discover in your draft league that you need to play with this number to get just the right balance of hitting and pitching in the upper end of the rankings.  You can change it as often as you like until you achieve the right balance and then, once you do, bookmark your results page in your browser.

Q.  Besides injuries or playing time, what other factors could alter your projection for a certain player during spring training?  I'm asking about things like stance or motion, apparent better physical shape or the opposite, maybe a new pitch or better spring training control?  Just curious.

A.  It actually takes a pretty extraordinary spring training, either positive or negative, for me to change the forecasts that late.  It has happened (I recall Jason Marquis a few years ago got a major downgrade just before the 2003 or 2002 season because of something I saw in spring training) and it will usually involve either (a) a player's physical shape being completely different than expected or (b) a hitter's contact rate showing something unusual.  I wrote about the contact rate aspect and spring training just before spring training last year in an essay called "Making Contact and Margins of Error" which you can read in the favorite essay section at our home page.

As for the physical size of a player, it's also usually a hitter who will get my attention in that respect though I don't believe in automatic upgrades or downgrade on changes in weight on their own.  Regardless, if you'd like to see some of my research on how weight affects hitting performance, I encourage you to check out the two articles called "Size as a Statistical Field" which are also at our home page, particularly part one of that two part series.

Unfortunately, I'm not smart enough to recognize how successful a new stance will be so I don't remember ever accounting for that, unless the results showed through statistically in spring performance and under a category like making contact.  I'm decent but unspectacular at detecting bat control problems but spring training is so short that it often isn't enough for me to make a change based on that alone.

From issue published March 15, 2006:

Q. I like your mathematical approach to projections but when it comes to confidence, I would recommend an adjustment away from 'high' and 'low' expectations for individual categories and begin to tackle a more useful output, which instead would consider high and low projections of a player's entire statistical profile - in other words, best or worst SEASONS a player is likely to have.  I think such an approach would be more meaningful/useful.

A.  There are several problems here.  First, how will we decide what is the best season?  If we simulate a hundred seasons, we need to be able to pick out the best and the worst to achieve what you're talking about.  What if in one season, our player hits .230 but steals 70 bases.  Check out Vince Coleman's 1986 season when he hit .232 with no home runs and 107 steals.  I don't know if I would have rated that as a good season and yet a fantasy leaguer might have loved it.  What if that same player in another simulation hits .260 with 10 home runs but only 45 steals?  Is that a better season?  I'm not sure which one is better either way but what I do feel is that I don't have the ability to scientifically look at two season lines and decide which one is the "better" season in a way that's useful to the reader, largely because for every reader, one category may be more important to them than another.  The fantasy leaguer who plays in a sim league may care more about OBP + SLG and other players might be in a "home runs only" league, which I know exist because I occasionally get questions about them.  The "home runs only" fantasy player doesn't benefit from me calling Coleman's best case season the one in which he hits no home runs and steals 107 bases.  The Sabermetric manager might look at his .301 OBP from that year and say that even with 107 steals, this is a bad season.

Moreover, and more importantly, I don't know how to project a best and worst season and do it with the quality where I would feel comfortable giving the numbers to readers.  For example, to achieve best and worst seasons, I would likely have to do a simulation that uses my projection as the base and then picks out the best and worst seasons the player achieved but the issue becomes how many simulations to do and whether the projection or base itself is the one I should be using.  What if it's wrong?  By using relative confidence levels in each category, as I already do now, I can actually look back to our historical results and ask the question: For this type of player with this track record of experience and consistency and data available, when I have in the past projected approximately this type of performance in this category, how wide is the range of performance to this relative level of confidence?

But to represent a seasonal line, I would have to find a way to take the projection and ask this question:  For this type of player with approximately this sort of projection in all categories, with this level of experience and this amount of data available and this track record of consistency, what is the typical best and worst seasonal performance achieved by this player X% of the time, with X being our confidence level?  The problem is that the question is so specific when you have to narrow the search to find players with similar projections across the entire batting or pitching line that there's no way to get a good sample size and thus a solid best and worst projection for the player.

Besides, I don't purport to publish "best" and "worst" performances in a category even.  Our relative confidence levels simply allow you to compare how likely Player A is to achieve between 10 and 30 steals as Player B is to achieve between 25 and 40 home runs.  There's nothing preventing either player from performing outside of the range in either category and they are not boundaries within which all or even most player performance will necessarily ultimately finish.

The short answer to your question is that for these main reasons, I can't think of a good way of doing this in the context of the way we do player forecasting.

From issue published April 5, 2006:

Q. I used your draft values last year and patiently waited for the bargains to arrive - and there were plenty - but ended up leaving about $30 on the table (total salary $260).  I could have upgraded one or two players with that extra money although it would be hard to anticipate where to do that earlier in the draft.  Do you ever have a few dollars left over in your drafts?  How big a deal is that and what do you do to prevent it, if anything?

A.  I think I've only once finished an auction and had more than a dollar or two left over (and that was due to a calculating error on my part during the auction) and my simple and short advice is this - Don't let it happen.  There's no good reason to hold money back and even it means you overpay according to strategy to get a little bit extra on your roster, every bit counts.  My own valuation method used by the fantasy domination forms has an aspect called "forced escalation" which I added a couple of years back to help with giving a bit more value to the top players and a little less to the bottom players.

During an auction, I almost exclusively use my methods described in "On Paying 80% of Projected Value" and usually I have no trouble using the money.  However, if this doesn't happen in your league and you find yourself with money left, chances are good that either (a) you went into the auction with a keeper list that wasn't ready to compete or (b) if you're not in a keeper league, you're against extremely good competition, owners who don't overpay for players.

If the second item is the reality in your league, then the only thing you can do when you realize you're in danger of having money left is to go ahead and approach paying a higher percentage of actual value.  I hate to think this could happen but basically, I'm saying it's better to pay 85% of projected value for a roster that costs you $260 than it is to pay 78% of projected value for a roster that costs you only $230.  In the first case, you've picked up about $305-$306 in value and in the second case, you've been smart on the percentage but ended up with only about $295 worth of value, which means you left the auction with an inferior team and a better salary to value ratio.

My only advice is if you're consistently finding yourself with money left in the auction year after year, you're either against extremely good competition and must adjust your percentage target accordingly, settling for a bit of a less favorable percentage (like 85%) or if you're in a keeper league, it's the key to you not running into this problem.  Ideally, you want a keeper list that's going to let you overpay for players during the draft as I explained in the essay "Keepers and Optimal Bidding."

One other approach here, and I've used this twice to win leagues but that's not enough to call it a tested theory, is to pay top dollar for your first two or three players and then kick in the 78% approach to assemble your whole roster.  You're not changing your target but it means you're doing what it takes to win the top few players and then you're settling for bargains from then on throughout the draft.  I really don't like this approach but I share it with you because I tried it and it did work both times where I both achieved a 78% target and used up all of my money in a very competitive league.  It's not tested enough for me to recommend it outright but I present it as another option.

Q. Let me please ask about auction strategy.  As I understand it, your 78% values per player are not so much to be viewed individually.  Instead, my goal going in the draft should be to buy me a team for 78% of their real value as a whole.  That means, I may be able to pay more than 78% value for an individual once I paid less than 78% of value for other individuals.  The trouble here is that the players I may need to pay extra for will most likely be brought out first or at least ahead of the players who become great bargains later.

A.  Yes, you have this right as I explained a bit above.  I suggest you check out the two essays I mentioned in the answer to the previous question as I continue to get many questions that are answered in those two columns.  You have an overall target and it's fine to overpay for some players if you can underpay for others, especially late in the draft.  If you feel confident that a specific player projected to be worth paying $10 for is going to be available to you later for $1, then you can account for that and use the $9 you expect to save earlier and on other players.

From issue published April 26, 2006:

Q. I was a bit confused by something that was in the "E-Book" version of your forecasts this past week.  Alex Rios's year-to-date batting average was listed as being higher than his OBP.  I was wondering if you could correct this typo and if it isn't, how it is possible that a player's OBP can be lower than his batting average?  Isn't that impossible?  Just wondering.

A.  I was amazed how many people noticed this and wrote to tell member support that there was a typo.  We always appreciate when readers catch us on something that slipped through the cracks but it may surprise readers to learn that this is not a typo and is actually correct.

In other words, it is theoretically possible for a player's batting average to be higher than his OBP.  Here's why - When a player hits a sacrifice fly, they are not charged with an at bat but they are charged with a plate appearance.  Thus, with sacrifice flies included in the denominator for the calculation of OBP, Alex Rios's OBP suffers from the 2 SF he had but it didn't hurt his batting average.  Uusally, a player will only have a higher OBP than batting average if they have almost no walks, which was the case with Rios when the previous updates were published as he had just 1 walk at the time.  Therefore, the listed OBP and batting average are actually correct as of the publishing date and more importantly, it is possible to have a higher batting average than OBP, though maintaining such an outcome over a full season would be unlikely, I expect, unless you almost never take a walk and get plenty of sacrifice flies.

From issue published May 10, 2006 (Associate Editor Josh Parks sat in for David Luciani):

Q. Should I always be trying to trade lowest projected remaining value for highest projected remaining value at this time of year? Are there exceptions?

A. Your overall goal in terms of your roster is to accumulate as much projected value as you can. So, you're on the right track with the idea of trading players where you get back more projected value than you give up but you can't stop there. You can think of that as sort of a strategic first principle, if you like, with emphasis on the word "first." If you are considering a trade the first thing to consider is whether the net result is more projected value for your team. The answer to that question determines how you proceed to evaluate the trade, as there are two main paths you can go down.

If the trade nets you a profit, the next thing you want to consider is perceived value. If you are going by our ranking sheets there are always many combinations of trades you can find that are favorable strictly from a projected value standpoint, but which would not necessarily be smart moves. Think about the player you are considering moving, where (or for how much) he went in your draft. Think about other players drafted in the same neighborhood, and what sort of year he's having so far, how established is he, what is his track record like? Once you actually have a handful of names you consider to be comparable you should have a sense of your player's perceived value.

Now do the same sort of thing for the player you are looking to acquire. Perceived value is just that, perception, so this is necessarily something you piece together by feel, not something you can calculate, but it's important to develop some sense of how to compare perceived value, particularly in your own league.

If there's a big imbalance against you in perceived value, you can then try to negotiate in favor of a deal that not only nets you a profit in projected value, but perhaps even in perceived value as well. You don't necessarily go into your thinking about perceived value with whomever your negotiating partner is. You communicate more in the sorts of names you bring up (i.e., "Well, I was thinking more along the lines of Player X"). If the imbalance is in your favor, and the deal doesn't create a big hole in your lineup, chances are you'd move forward with it.

This is, of course, a high-level view of some of the larger moving parts when it comes to thinking about a trade, but it does outline one path by which you can arrive at deals that should tend to help your team.

You asked about exceptions, and in fact there are. If somebody offers you a deal that doesn't net you a profit in projected value, that shouldn't be the end of the story either, it should just send you down a different path.

Let's say somebody offers you Chris Carpenter for Jason Bay. Let's say that going strictly by the ranking sheets, Bay has more projected value, so you're inclined to reject the deal. But let's say you then look at the deal from a perception standpoint, and you find that Carpenter's value was significantly higher. You'd want to consider it, with the an eye toward perhaps then trading the higher perceived value for even more projected value.

So while the goal is to maximize projected value, it's not as simple as trading low projected value for high projected value. To get the most out of your roster you have to make perceived value an essential part of any trade decision.

From issue published June 14, 2006:

Q. I made a trade in my league that the ranking sheets at your website projected would be very favorable to me.  When we announced the trade, the commissioner said that the trade was suspect and it went to a league-wide vote.  The majority of the league ruled against me and the trade was overturned.  This question isn't about the players involved but really, I want to see if you have any advice for me in future trades so this doesn't happen.  I really think I would have been getting the better of this deal performance-wise and yet, even I have to admit that it looks really lopsided against me.  What am I missing here?

A.  You know, I saw someone discussing a similar topic in our free discussion forum and while I didn't comment, a thought came to mind about this where even I've changed my own mind compared to my thinking of a few years back.  Several years ago, I wrote an article called "On Fantasy Trading Rules" about why I had to exit a league when I had a trade overturned that would have turned out to be particularly lucky for me had it been allowed to go through.  In hindsight, I realize that I was wrong then and that's because I wasn't obeying my own strategy regarding perception.

That is, I think I've gotten much stronger as a fantasy player over the years because, especially since about 1998 or so, I don't just look for trades that benefit me.  In fact, it is the rare case indeed where that criteria alone will clinch a deal.  Rather, I want a trade that both benefits me and appears to benefit me.  Here's why.  If the majority of people in my league felt that should I have gotten more in a deal, then I probably could have gotten more in a deal and it means that I am underselling my assets simply because I believe I'm getting better forecasted value.  Perceived value is a separate and real entity that is important to have on your roster because trading power generally comes from the perceptions about your players more than it comes from forecasted value.  Having trading power is crucial to success and so you not only want to be adding players who will help you in terms of their forecasted performance but you also want to be adding players who give you trading flexibility by virtue of their long-term ability to be moved for even more value to another team.

Let me make an ridiculous example to illustrate the point.  Let's say that you get struck by a bolt of lightning and become psychic and somehow could know with absolute certainty that from tomorrow until the end of the season, that little-known Sal Fasano would hit 40 home runs the rest of the way.  Let's say on your roster that you had Alfonso Soriano and Ryan Howard and let's say that your psychic ability also told you that neither one would hit another home run from now until the end of the season.  Does that mean you should trade Soriano and Howard for Fasano?  Of course not.  The reason is obvious.  Even if this extreme example were true, you could get so much more for Soriano and Howard than just Sal Fasano.  You'd be maximizing projected value but not perceived value when you closed the deal.

The essence of success in fantasy baseball is not only adding real value but also maximizing the return on each investment you make.  Every player in one way or another is an investment by your team.  Whether you invested a round two pick, a $40 draft day bid or the trading of another player previously, you have made an investment in the player and what he will do for your team in the long run, either by virtue of his actual performance or by what he will help you acquire.  To focus only on the performance and not the perception aspect is to miss half of the fantasy game.  It doesn't mean that you have to force equal perceived value on every trade.  Certainly, it's a lot harder to convince someone to trade with you if they feel they're given up too much.  But you can push it to the absolute limit.  Get things to the point where your opponent things he's giving you 5% less than he's getting rather than 30% less and you'll be truly getting the most out of the assets on your team.  Little edges like this add up to big advantages in the long run.

So, in terms of your league overturning the trade, and I never expected I'd be saying this as I remain opposed to league veto rules in most cases, be glad that they overturned it and go out there and get both lots of projected value and lots of perceived value and you'll be glad you did.  You didn't name the players involved or otherwise I would have given you more feedback on what you might have done to get more in your particular deal.

Q. One of my tricks for success has been to always draft players in the final year of their contract.  They always play better in the last year because they want to play well to get a new contract - like Carlos Beltran, Adrian Beltre, etc...  Why don't these guys get bid up more in the drafts?  Is it just that not enough people know about the strategy or is it because people won't pay more for a player than they've ever seen them be worth before?

A. I get this question every few years and once it starts to become a common question again, I have to offer a public response.  Believe it or not, I don't subscribe to this theory.  I used to do an annual study on this issue every year or two up until 2002 when I finally realized that no year was proving to be the exception.  On average, players do approximately the same in their final year prior to free agency then they do in any other year.  No doubt, many players have career years in the year leading into free agency but those are the ones we notice.  If you want to see the final study I did on this issue in 2002 (a study that MLB.com also had me write about for them at the time), you can read it in the archives at the website under an essay called "The Potential Free Agent System."

Every time I draw this or any other such analysis to the attention of readers, I get many responses from people who try to argue otherwise but they always quote only the portion of results that supports the theory (i.e. the names of players who did well in their final year before free agency) rather than high level evidence that looks at the entire population.  When I first researched this topic, I actually went in admittedly a bit biased, wanting to believe that it worked and would give me an advantage over my fantasy opponents.

Unfortunately, I now realize that to believe the theory means that I have to not only see data and reject the implications of that data, but it also means that I must accept that players actually don't try as hard as they could in years other than their final year prior to free agency and then suddenly try or work harder in that final year before a contract is played.  Unfortunately, we have to keep in mind that a player is actually playing for his contract with each season, particularly those three years prior to free agency when he is subject to the offers and tangles of possible arbitration with his club.  His numbers then carry plenty of weight in terms of how much he will be paid and so perhaps a new theory, one I haven't investigated in great detail, would be to look at how a player performs in his final year before his first arbitration-eligibility.  Anyway, I just don't believe this theory.  I wanted to because I'm always looking for an edge but most evidence convinces me that while there will be players in the potential free agent class who will have career years, we just don't know which of them it will be and for every such over-performer, there will be an underperformer that negates the utility of the strategy.

To answer the last part of your question, it's possible that no one else bids up these players because they don't believe the theory either.  If you're one of the readers who remains convinced that it's a good strategy, then by all means keep using the system.  I can't recommend it, though. 

From issue published July 26, 2006:

Q. This is a multi-part question - First, a few years back, I remember you published something like a formula for projecting real team win loss records. Can you publish it again and using it, do you think the Detroit win/loss record is really in line with how they've really performed on the field or have they been have they been lucky? Also, with 20/20 hindsight now, can you look back and see something in the stats that were available to everyone [like three year averages maybe?] that should have been able to project the Tigers to be so good?

A. Well, I certainly didn't predict the Tigers to do this well, having published at the website back at the end of spring training that I thought they would go 80-82 this year, still a 9 game projected improvement over last year but nowhere close to the success they've had.

Let's see first if their win/loss record is a realistic representation of their team performance this year or if they've just been lucky. In this case, I'll look at games completed through this past Friday, when the Tigers' record stood at 65-31. As of that date, they had scored 498 runs and allowed 364 runs.

If you use the modern Bill James' "Pythagorean formula" for projecting wins (being as follows: projected winning percent = (runs scored ^ 1.82) / ((runs scored^1.82)+(runs allowed^1.82)), you end up with a projected win loss record of 61-35.

My own formula for projecting win percentage, which I make no claims to be more or less accurate than the James' one, is as follows:

Projected WPCT = .495+(.083*dpg)+(.00162*dpg^2)+(-.0000706*dpg^3) where dpg = the number of runs more that a team scores per game than its opponent.

The Tigers have outscored opponents 134 runs over 96 games which equals a dpg of 1.40 (i.e. the Tigers are outscoring opponents an average of 1.40 runs per game). Their projected winning percentage under this formula should be:

Projected WPCT = .495+(.083*1.4)+(.00162*1.4^2)+(-.0000706*1.4^3) = .614

In other words, my own projected team winning percentage formula produces a projected record of 59 wins and 37 losses and that means both formulas figure that the Tigers have probably won anywhere between 4 and 6 games more than they should have, given their actual runs scored and runs allowed stats this season. In other words, they're close and they've certainly earned their win/loss record without extraordinarily good luck but they're run scoring / runs allowed totals do imply a slightly lower record than they've had.

At the individual player level, one error that's easy to make when looking a team's performance is to identify the team as having improved over last year by singling out totals from the hitting or pitching side of a year ago. Often, the improvement isn't because of real improvement of the players but rather replacement of the players by free agency, the minor leagues, trade or by simply giving playing time to players who never had as much before and taking away playing time from players who didn't perform.

To answer the part of your question about whether I or anyone should have seen this 2006 season coming, I thought it would be fun to compare how the Tigers have done this year in relation to the history of every player they've had play this year. This will tell us whether the failure to project this season was because we didn't accurately project who would play or whether the failure is a result of not projecting how well each would play. Of course, we know that both are always contributing factors in a missed team forecast but knowing how much of each is involved helps us improve future forecasts.

What I've done here, then, is weighted all the historical stats in line with how much playing time each player has received this season. So, a player who has had only 1 at bat in 2006 is weighted for only 1 at bat in terms of how much we care about what he did in 2005 and a player who has had 200 at bats in 2006 gets a weight of 200 at bats in terms of our interest in his history (in fact, I'm using "outs made" rather than "at bats" but you get the idea). I do a similar adjustment on the pitching side for innings pitched.

This chart really reveals the improvement of the team overall. In this case, I've used all play completed through July 20. To make the data more readable, I've scaled the historical results to the number of outs made (by hitters) or innings pitched (by pitchers) the Tigers have had this year:

Let's start with the offense, all scaled to the same number of outs made by hitters:

As you can see, believe it or not, this top ten offense would have actually been entirely predictable if we just could have known how much each player would play in 2006. Granted, the players are showing a little more power than they have historically but if you look at the two year average (my preference for most hitting categories), you see that essentially the players on the Tigers are actually performing entirely in line with their 2004-05 averages. Again, the 2004-05 line above is all data that was available to us before this season with the exception of knowing how much each player would play in 2006 but the key is that the historical stats are weighted based on playing time in 2006. If we had known how much each player would play, we would have been projecting around 450 runs scored after 93 games played or about 40 fewer than the Tigers have actually scored this year. For what it's worth, the Tigers only have one player on the hitting side who had not appeared in the majors in 2005 and that's Jack Hannahan and his 9 at bats.

On the pitching side, we start with three pitchers on the Tigers who have made significant contributions this year that we simply had no historical data for because they hadn't appeared in the majors in 2005. So let's begin by setting aside those three (Zach Miner, Joel Zumaya and Jordan Tata combined) as being unpredictable based on historical multi-year average stats because we didn't have any big league data. In fact, I actually use my own minor league translations when making forecasts but they wouldn't have been much help here anyway as well as these three have pitched:

Looking at the rest of the pitching staff, here's how they've performed this year and how their historical stats would have implied they would perform, again weighted by 2006 playing time, in this case innings pitched:

Unlike the hitters, what we see here is a real and complete improvement no matter what historical stats you use. In fact, if you use the most optimistic of historical data sets (2005), you still see an improvement of a full run on ERA on the pitchers who had appeared in 2005 and the Tigers also got the benefit of over 100 innings so far from that 3.01 ERA group of pitchers making their debut.

In summary, if you had told me exactly how much each player would play and using just historical two year averages, we might have reasonably projected the Tigers to score about 450 runs but we never would have been able to guess how good this pitching staff would be. Now, of course, the question is whether the pitching staff can maintain this pace then as the likes of Kenny Rogers, Jeremy Bonderman and Justin Verlander head toward career years.  As I write this, Kenny Rogers' is seeing his ERA come back to reality in a single game...

From issue published August 16, 2006:

Q. I appreciate you sharing your own "wins formula" for projecting team wins based on runs scored and runs allowed but it actually wasn't the one I remembered you publishing before.  There was another much simpler formula you had - can you please tell me what that one was?

A.  Sure, though I don't use the simplified version as often now and haven't since 2002 or 2003 I think.  The simplified wins formula is quite easy to use and actually yields decent results if you're in a rush.  That formula is:

Projected Winning Percentage = .500 + (.103 * run differential per game)

Run differential is simply the number of runs more that a team has scored than its opponent, per game and it can be a negative number if a team is being outscored by opponents.

So, Detroit's projected win total based on games played as of August 13, 2006 would be:

Projected Winning Percentage = .500 + (.103 * (137 runs scored more than opponents / 117 games played )) = .621 compared to their actual winning percentage of .650