taking away the right cross

dave studeman's article yesterday at hardball times added some insight to the recent flurry of posts about groundball/flyball ratios. to summarize what we've learned so far:

• cardinal diaspora established that the cardinals' pitchers tend to become more groundball oriented under dunc/tony. see posts on stl's starting pitchers, the relievers, and ryan's conclusions.
  
• baseball musings established that above-average gb pitchers, taken as a group, have slightly better-than-average eras -- weak correlation, but nonetheless real.
  
• the 26th man wondered if gb-oriented pitchers were an undervalued commodity and found that they're not; their cost is consistent with the overall market for pitchers.
  
• i looked at some data which very superficially suggest that teams with power-pitching staffs have an advantage in the postseason over teams with put-it-in-play staffs.

studes' article crunches batted-ball data and offers this, as one of several findings: "Strikeout pitchers tend to give up lower-impact fly balls." that is, the fly balls they give up tend to result in fewer runs than fly balls yielded by other pitchers. why is that significant for a staff as groundballish as stl's? because fly balls, for fairly obvious reasons, carry much heavier run-scoring potential than ground balls, and have a much broader range of potential outcomes. the vast majority of grounders result in either outs (negative run-scoring potential) or singles (small positive run-scoring potential); fly ball results, by contrast, range from outs (negative potential) to doubles, triples, and home runs (high positive potential).

studes' batted-ball leaderboards article, posted in november, helps to illustrate this point. from 2002 through 2004, the most productive flyball hitters -- guys like bonds, edmonds, manny, sammy, a-rod -- generated an average of about .25 runs per fly ball. by contrast, the most productive groundball hitters -- e.g. ichiro, chone figgins, kenny lofton -- generated about -0.25 runs per grounder. that's right -- even the most dangerous ground ball hitters hurt their offense when they hit a grounder. fairly obvious why: most grounders are outs (negative run value), and the ones that are hits carry only a small positive value; average them all out and the result is negative.

a pitcher's real margin for gain, then, lies in controlling the flyball hitters. if, as a pitching staff, you hold edmonds pujols et al to .05 runs per fly instead of 0.25, you're saving runs in bulk over the course of a series.

now return to studes' article from yesterday and take a look at the fifth graph, labeled "batted ball relative run values -- pitchers". the left-most column ("per of") refers to outfield flies; the one in the middle ("per gb"), to ground balls. see how much more compressed the range of values is for groundballs? the spread between the most efficient groundball pitcher in the league and the least efficient is only about 0.1 runs/grounder, and the spread between the most efficient guy and the average guy is even less -- about 0.05 runs/grounder. being way above average here doesn't save you many runs. the corresponding spreads for flyballs are about twice as high, so if you're way above average in this category your run savings is much more meaningful. again, that's where the margin for gain is.

think of ground balls as jabs and fly balls as right crosses. there's only so much advantage to be gained by neutralizing the other guy's jab, because he can still hurt you badly if he lands a right cross with full impact. conversely, if you neutralize the other guy's right cross, there's only so much damage he can inflict with his jab.

so when studes says, "Strikeout pitchers tend to give up lower-impact fly balls," what he means is: strikeout pitchers tend to take away the right cross. that's exactly what teams have done to the cardinals the last couple of postseasons, no? our power hitters pop up or fly out to the track instead of driving it into the gap or over the wall, and we're left with nothing but the jab.

"Another truism," studes writes, "is that when groundball pitchers yield outfield flies, they tend to have more run value." oh, great; so even if stl's pitchers induce 28 groundballs and 12 flyballs in a given ballgame, one or two of those dozen flies is apt to break the cardinals' jaw . . . . . we've seen it happen, haven't we?

i don't know exactly what data studes is basing his assertions on, but i've e-mailed him to try to find out. also to see if he can direct me to some sources (i suspect the THT 2006 annual is one on'm) so i can nose around on my own. i may take a look at the cards' postseasons since 2000 and calculate their run values per flyball and per grounder, to see if empirical data support studes' assertions; i'm pretty sure there's sufficient play-by-play data out there in the public domain to support such a study. but it'll be a slog, and i won't have answers for a while.

so for now, just toss this into the mix, and we'll try to get it sorted out by game 1 of the nlds.

p.s.: one potentially encouraging piece of news: according to studes, one of the best relievers in the game at minimizing the impact of flyballs is felix rodriguez. . . . .