clutch hitting --- ah, the eternal (or should i say infernal?) question.

not that many years ago, bill james studied clutch-hitting ability exhaustively and declared it a mere illusion. doesn't really exist, he said with certainty; a player might gain a reputation as a clutch hitter via a few well-timed and -remembered big hits, but there's no evidence that any player, over a large sample size, can reliably raise his game with wins and losses on the line. rather, james concluded, players' clutch production mirrors their non-clutch production over the long haul; a guy's abilities don't change simply because the game context changes.

james has spent the last 10 or 15 years tiptoeing away from this finding. with better statistics and faster computers, sabermetricians are now better able to isolate and study clutch situations --- and they're finding that clutch-hitting ability does exist, at least in a limited number of cases and within a limited set of parameters. most recently and persuasively, andy dolphin asserted in The Book that clutch hitting is a bona fide skill --- rare, but very real. in all but a handful of cases, james' axiom still holds true --- for most players clutch hitting comes and goes randomly, regressing to the mean over time. every string of big clutch hits a player delivers will be canceled out by a proportional string of clutch outs. david eckstein serves as the prime example of this principle: no cardinal was clutcher than he in 2005, but in 2006 he couldn't buy a big hit --- not until the last two games of the world series, anyway. albert pujols might serve as another example. it's not well remembered anymore, but one of the major storylines for most of 2005 was pujols' pattern of coming up empty in game situations. (see this post from august 2005 for an example.) of course, game 5 of the '05 nlcs forever abolished our memories of albert's regular-season unclutchness, and his incredible performance in the clutch throughout 2006 cemented his legend. he's now a Certified Clutch Hitter; has a ring to prove it and everything.

but should we rely on him to produce in the clutch again in 2007? only at our peril.

last season, fangraphs' david appelman began tracking a gadget statistic called "clutchiness," which derives from this post by tangotiger at The Book blog. the best definition of the principle comes from the Clutchiness blog:

fangraphs calculates clutchiness via a mathematical formula that you can find here, at the very bottom of the page. here are the cardinals' 2006 leaders in clutchiness (abbreviated "CL" in the table). the values expressed below correspond to wins:

first things first: you'll note that the clutchiness number doesn't always equal the difference between WPA value and OPS value. that's because there is a multiplier involved, called leverage index, which is not worth the trouble of explaining in depth. the short explanation is that leverage index measures each player's opportunity to impact win probability --- his clutch at-bats as a proportion of his overall at-bats, roughly speaking. you can't really fault a guy for having a low clutch value if he rarely comes up in clutch situations.

continue reading after the jump . . . .

to nobody's surprise, pujols led the majors in this category in 2006 --- albert was so good in the clutch that he contributed 2.67 more wins than his raw statistics would have suggested. this is all the more incredible when you consider that teams routinely pitched around him in clutch situations, which artificially lowered his win-probability score. fangraphs' clutchiness figures only go back to 2002; in that five-year period, only one player has recorded a higher number than pujols did last year (david ortiz posted a 3.72 clutchiness score in 2005). but just to show you that clutch performance can be fleeting, last year was the only time in five years that pujols has posted a positive clutchiness figure:

over his career, pujols has been slightly less impactful in the clutch than in non-clutch situations. note that he's still incredibly valuable overall; his WPAs of 5.76 in 2004 and 5.72 in 2003 were among the best in the game for those years, so we shouldn't conclude from his poor clutchiness numbers that pujols' great stats don't translate into wins. on the contrary, they translate into tons of wins. but until last season, albert apparently didn't produce at his usual rate in situations where the game really hung in the balance, when a hit might have turned a loss into a win.

now go back up to the table listing the cards' 2006 clutchiness leaders. look at how prevalent the postseason heroes are on this list (which, remember, only accounts for the regular season). pujols had the decisive hits in the first 2 games of october, getting the cards off on the right foot; spiezio delivered two of the biggest hits of 2006 (the triples vs the brewers in the 160th game of the year and vs the mets in game 2 of the nlcs); no need to remind you of the game-winning homers by molina and taguchi at shea. perhaps the latter two clouts weren't so shocking after all; taguchi and molina had been raising their games in the clutch all season long. and it's even better than that: taguchi and molina are the only st louis players whose clutchiness score has been in the black three seasons in a row:

TAGUCHI

MOLINA

here, we find the opposite of what we said about pujols above. these "clutch" players are still net losers, as indicated by their negative WPA. over the course of a given season, they generally contribute more to losing than to winning. but they also very clearly, and very consistenly, get the most out of their limited skills when the game is in the balance. if they're below-average hitters overall, maybe they perform like average hitters in the clutch. follow?

we can substantiate that impression with one of the more traditional measures of clutch performance: hitting with runners in scoring position, or RISP. the following are career totals:

check out molina: his isolated power w/ RISP dips nearly 30 points, but his batting avg goes up by 58 points and his avg on balls in play (BABIP) increases by 67 points. taguchi shows a similar tendency --- iso power dips from .118 to .097, while his batting avg increases by 60 points and his BABIP soars by almost 90 points. these stats lead me to believe that we're not really measuring clutch ability in these figures; rather, we're measuring (one of my favorite concepts) situational intelligence. i think that with men in scoring position, these players cut down on their swings and simply try to put the ball in play and drive in the runners. they hit for less power but get more line drives, as reflected in the higher batting avgs and BABIPs. and the impact of those rbi hits --- higher win probability --- gets captured in the clutchiness stat.

there's a lot more to say about this, but i've gone on more than long enough; spent an extra 40 minutes on this post that i was supposed to have spent on the stationary bike. another lump of muscle mass sacrificed at the altar of baseball statistics . . . . please don't tell my insurance agent about this.