Julien's Faster than Light Blog
Jump to top
Friday, April 28, 2006
frequently asked question
the frequently asked question is "what are all these weird stats you use?"
on june 19, 2003 i invented the wal con pow system. the wal con pow system breaks a player's performance down into component skills, then uses those skills to predict future performance. it turns out that different skills peak at different ages, and different skill profiles imply different career lengths.
the current formulas are:
wal = w/(aab + w)
con = (aab - k)/aab
pow = xbh/(aab - k)
where w = bb + hbp, aab = ab + sf, and xbh = extra-base hits.
aab stands for "adjusted at-bats". for pitchers, the numbers are a little different because i haven't been able to get at-bat or extra-base hit data. i've been using ip*3 + h for aab, which of course also includes sacrifice bunts, double plays, and caught stealings. so the wal's will be a little lower than they should be, and the con's will be a little high. instead of pow, i've been using hrp, which stands for "home run percentage." the formula is hr/(aab - k).
con tends to peak around age 25. pow peaks at age 28. wal peaks a few years after pow, as players learn that pitchers are afraid to throw strikes.
ip*3 doesn't include errors, so that mitigates things. but were getting too far afield.
the point here is to talk about the value of the wal con pow system.
so what is the value of the wal con pow system? for one, it makes it a lot easier to tell who's going to make it. contact percentage, it turns out, is a good measure of how much trouble a player will have adjusting to a new level of competition. a player with high contact will have very little drop-off, while a player with low contact will have a lot of work to do.
for pitchers, it's a little different. con is an important measure of how dominant a pitcher will be, but wal is more indicative of the ability to move up, because hitters are more selective at higher levels.
the other thing the system is good for is matchups. players with different skills will have different strengths and weaknesses in the batter-pitcher matchup. an example of how this applies can be found here.
the investigation into matchups has made me realize that run-based player analysis is fatally flawed. traditional sabermetric measures include the assumption that a team's win rate can be approximated by its runs scored and runs allowed rates. this assumption causes skills such as defense, speed, and contact to be undervalued, while patience, power, and power are overvalued. i couldn't think of a third thing to put in that list.
but power is overrated in players that don't make contact. those players will have little to contribute against dominant pitchers, but will pad their numbers against soft-tossers. you can see how an analysis based on runs would skew one's perception of such players. examples abound throughout the site.
in closing, thanks for visiting. feel free to use the comments or write to me.
Comments:
Post a Comment
