More Masochism

I don't often write about how the dollar values I use are derived because I don't want to give my readers an ice cream headache. But my post on how I do derive my formulas has generated a lot of interest. Eugene asks:

I would like to know if you derive your ERA/Ratio/BAvg numbers based upon league averages or rotisserie league averages and where you get the IP and AB factors that you include in those formulae?

I use Rotisserie League averages and not league averages. Since Eugene asked the question, he presumably knows the difference, but for the rest of my audience I'll explain.

Alex Patton's formulas take a midpoint number in ERA/WHIP/BA and determine that this number is the midpoint for player value in the category. For example, if your midpoint is a .270 batting average, a hitter who hits .270 earns $0 in the batting average. Anyone who hits below .270 in batting average loses money and anyone who hits above .270 earns money. The idea is that the earnings in the category will be $0.

Incidentally, this is not a universal precept. Some analysts believe that you should take a number below the midpoint and use this as your baseline. The primary difference here is that your earnings in the category will be greater than $0. If anyone is interested in how this works, I can try to explain, but that explanation is beyond the scope of this post.

If you are using a Rotisserie League averages, you are using a pool of players purchased by a typical Rotisserie League. You can derive this pool of players any way you like, but hopefully you are using a population of players that is realistic. Adding Desmond Jennings to your hypothetical player pool, for example, would be realistic if your league allows you to buy minor leaguers at auction. Adding Fred Lewis, on the other hand, would not be realistic, since it is extremely doubtful he was purchased in an A.L. Rotisserie-style auction in late March or early April 2010, even if your league did allow you to purchased players from the National League.

Using league averages means that you are taking the league's ERA/WHIP/BA and spinning off a Rotisserie League's ERA/WHIP/BA from that number. With this method, it is irrelevant who was or wasn't in your draft population; it only matters what the league did as a whole that year. In other words, the $ that all of the American League pitchers earned in ERA in 2010 should be the same as it was in 2009, in 2006, or 2001.

I use Rotisserie League averages because I'm interested in measuring how the auction population does year over year. The $3,120 or $3,380 dollar amounts are derived from what we spend at auction every year, not from what the player population as a whole earns. I don't believe that the ERA$ has to be the same for the entire American League or National League year-over-year because in some years the free agents who come into the player pool are better than in other years, and the earnings of the total player pool should (in my opinion) reflect this.

The second part of Eugene's question was where do I get the IP/AB factors that I include in the formulas.

The 5,500 AB and 1,200 IP I use in the formulas came directly from Patton's software. The same is true of the .0016 (BA), .011 (WHIP) and .07 (ERA) multipliers that I use to calculate these values. I'll be the first to admit that these numbers are discretionary. If you want a player's contributions in the qualitative categories to be worth more or less you can fiddle with these numbers all you like, as long as you make sure that you are still consistent with your rationale: whether it involves league averages or Rotisserie averages.