"Mr. Madison, what you've just said is one of the most insanely idiotic things I have ever heard. At no point in your rambling, incoherent response were you even close to anything that could be considered a rational thought. Everyone in this room is now dumber for having listened to it. I award you no points, and may God have mercy on your soul"

That line could very well be what you think when you finish reading this and the upcoming posts. In the following posts I'm going to layout my thoughts on drafting strategy.  This particular post is going to get into the construction of a proper statistic to compare hitters within position, across position, and then incorporate a comparison of hitters vs pitchers.  Let's start with a few examples, and let's take out all uncertainty by focusing on actual 2009 stats instead of any projections for 2010.    As usual, I'm assuming you're playing in a standard RotoLeague with Rs/HRs/RBIs/SBs/AVG/  and on the pitching side (Wins/SVs/Ks/ERA/WHIP)


Question 1: Who was better Dustin Pedroia or Brandon Phillips: 

 Let's throw out some options,  1)  Brandon Phillips was better in HRs, SBs, and RBIs and Pedroia was better in Runs and Average, but that doesn't really help since it doesn't take into account how much better.  2) One of the metrics I like (but don't use) is to look at each player and calculate what percentage each player gives you of a certain target. I like choosing the second best finish in each category for my target.  For my league, that was 914/252/901/154/286.  This changes the stats to 

   


(I divided average by 10 under the assumption you have 10 players in your lineup on average).   If you average those you get 10% for DP and 11% for BP.  So, on average either player would get you about 10% of the way towards hitting second place in each category. I think this is fairly robust and you could rank players well by that average stat.  In fact, here is the standings for Second Base, ranked in that manor.

However it does have some weaknesses.  The first being that the target used is specific to my league and may not reflect your league or even my league next year.  The second and most important, is it doesn't take into account SCARCITY.  I put it in caps because if you spend any time reading fantasy baseball forums, people take scarcity to an extreme, to the point that people pass on Albert Pujols because first base is a "deep position".  That's just crazy talk but it should still be looked at. 


 I view two different types of scarcity that can affect a particular player's value 1)  The scarcity of productive players at a position (i.e. After Hanley Rameriz, all the other shortstops suck) and 2) The scarcity of players that supply meaningful stats in a category (i.e. No one steals bases anymore so I better draft Jacoby Ellsbury).    To take into account these two factors, you need to compare a player's stats, not to some target, but rather to the pool of players you'll be forced to choose from if you choose one over the other (i.e. how much is each player worth versus the opportunity cost of not picking them).  To make this comparison, I developed a stat I loosely call the Fantasy Z-Score (which, after doing some research, the guys at RotoRebel also use but I swear I thought of it separately, I feel like the other guy that developed Calculus).  The calculation is a little complex but I think it has a bunch of small benefits.  First calculate the player's production in a particular category (I like to use projections from Steamer, ZIPS, Cairo and Clay Davenport, with playing time projections from Fangraphs. I then try to equalize ABs for injury prone players since I don't do H2H, and I equalize for low IP forecasted players since I don't compete in GS leagues.  You do this by adding replacement level stats to their projected stats. Next subtract the average production of players for the entire pool of players likely to be drafted. Divide that by the standard deviation of the production in that category for the pool of players likely to be drafted. This is the z-score for that player by category. For rate stats like batting average, era and whip, multiply by the AB and IP divided by the average AB and IP for batters and pitchers likely to be drafted.  You can then average those category z-scores together to get one score (or total them).  Then split the pool into positions (placing each player in his most shallow eligible position). To do this split you also need to allocate the players into your utility slots (MI, CI, and Utility).  Assume that each player in your league drafts the highest available Z-Score for each position (based on position limitation).  Once you have that,  then need to determine each player's Z-Score Total above the lowest level player (replacement level).  This takes into account positional scarcity and the marginal z-score is what you should rank and value player's based on.  For auctions, divide the entire money available to purchase players (after keepers are subtracted) by the sum all the positive marginal z-scores available to be drafted.  This is what you pay per marginal z-score.  Multiply that by each player's marginal z-score for his value.  I also tend to due this separately for hitters and pitchers based on a standard 70/30 split in my leagues for hitters vs pitchers. If I'm doing a snake draft I like to do the above and then rank the players by that auction value. 1) It weights down the pitchers, which is typical and perhaps reflects risk in pitchers vs hitters and 2) it keeps it clear as to what reaching / overspending really means. It might just mean a few dollars (or it might mean $15).  That is all pretty confusing, and if you duplicate the steps you'll see that some are iterative (you need to attempt, take the output and re-run the analysis, like for finding replacement level players).  If anyone is interested, I'll provide the How-to on this in a much longer, more detailed piece.  I'll also provide my rankings for a few league formats once all the projections systems are out.