THE BOOK--Playing The Percentages In Baseball

I’m kind of surprised at what seems to be an increased attention to ‘clutchiness’ these days. I guess it’s a bit of “have data, will use it” (reference to Palladin, for you old time TV watchers). It’s certainly a legit part of a “descriptive” analysis, but it’s pretty much the last stat I’ll pay attention to on the Fangraphs player pages.

In my view, it’s only by confronting it head-on, that you can expose it for what it is.  By calling it Clutchiness, as opposed to Clutch, you are giving it a double-meaning (see: Truthiness).  No player goes from being worst in the league in something to best in the league in something year-to-year.  Yet, here we see Pujols, going from -2 wins to +2 wins in Clutchiness.

Descriptively, clutchiness shows whatever everyone else sees: David Ortiz is God of all that is clutch.  But, Pujols’ clutchiness shows what we don’t see so readily: there is huge variation in clutchy scores, that whatever it is we think we see, is almost certainly the product of good timing.  (Clutch exists, but it is hidden behind a mound of noise.  It’s not so easy to spot.)

Patrick Roy is hockey’s versio of David Ortiz:
http://www.thedailycamera.com/sports/hockey/avs02/may/28saside.html

Roy improved his overtime record in the playoffs to 40-16 while Hasek’s fell to 12-11.

That’s as of May 2002.  Roy and Hasek would be the baseball equivalent of Clemens and Maddux in terms of talent.  And that’s being nice to Clemens and Maddux.

And in one of the most important games of his life, against the Redwings, in Game 7, a couple of years ago, Roy had the worst game ever.  It seems when you are hot, you are not.... until you are not.

If you want to know why a team has won 1 game (or 100 games), clutchiness is useful, instead of just looking at how many games they should have won (pythagorean).

Yes, that’s correct.  Clutchiness is descriptive, and explains what is happening, just like your brokerage account explains what is happening with your money. 

Neither shows that you are putting an effort into being better.  But, the Redsox and your spouse are both happy with the results.

Well, the reason people look at statistics throughout the year is because of their descriptiveness.

If you are only interested in the skill that statistics represent, you shouldn’t even bother looking at statistics during the year because these statistics will only give you a distorted view of a player’s skill, if they’re different from what would be expected.

A lot of statistics are skill descriptive (OBP, SLG, BP’s PX etc.) because they carry through year-to-year.  Others are event descriptive because they evidence something that happened, but not something found to be statistically meaningful for a particular player over time.

So, skill descriptive statistics are relevant as they are useful for predictions or to assess what the player is capable of producing---the event descriptive stats just say ‘this’ is what happened in ‘that’ circusmstance without giving any indication it will happen (in a probability range) again.