fPAA

In roto baseball, each category is essentially equally valued, but the distributions are all over the place. I set out to develop a fantasy rating method similar to the linear weights system used to calculate wOBA, and have calculated things assuming a typical Yahoo! 12-team 5×5 rotisserie, mixed league. The concept is not new, and very similar to Razzball’s Point Shares system.

What is “average”?

This is a critical question in this type of calculation; after poring over fantasy league data from previous years, I found it problematic to compare total statistics from sequential seasons, since overall offense is very different year to year (*ahem* steroids). So the model is based just on 2009 results.

I defined average in two ways, because we are interested both in overall production as well as production rate. That is to say, a player who hits 40 HR in 600 AB is more valuable than one who hits 20 HR in 300 AB. However, that 20 HR player is also more valuable than one who hits 25 HR in 600 AB, because you can play someone else for 300 AB and add their production together. I calculate relative fantasy values for each category in two separate ways (based on overall numbers and then on production rate), then average the values together to give one single number which represents the average number of roto points you can expect to gain (or lose) relative to the average performance of a standard 12-team, 5×5 roto league (think Yahoo! public leagues).

I assumed a roster of 9 productive position players, 6 starting pitchers (~1000 IP) and 4 relievers (~250 IP) which would make up the 1250 IP limit.

What happened to PRS and fPAR?

After lots of wrangling with my initial Percentile Ranking System (PRS), I believe that it is inherently flawed because talent in the Major Leagues is not evenly distributed. It is solid when it comes to valuing players in the main clump of “average” talents, but fails at the extremes of some stats, undervaluing the players with the best stolen bases, for example. This caused problems with player valuation in the top rounds, which is unacceptable if you want to win your fantasy league.

fPAR is very similar to fPAA, but suffers from at least two flaws: the determination of “replacement level” is hard to grasp compared to “average”, and also that replacement always changes based on league and roster size. While there are some factors which need to be adjusted in fPAA to compensate for these factors, the impact is far less when we view a player in relation to the average performance for a league.