Thursday, February 10, 2005

deranged: eckstein, chapter 5

not to sound like chicken little . . . but the more i delve into david eckstein’s fielding stats, the more i find not to like.

this despite some reassuring words from black hawk waterloo, author of the excellent angels blog chronicles of the lads. per the ongoing discussion of eckstein herein (see posts of feb 8, feb 6, feb 1, and jan 9), waterloo writes: "I was very surprised to see Eckstein fare so poorly by David Pinto's method [ie, Probability Method of Range]. Most others have him right around average last year, and he looks pretty good to the naked eye (which of course can be deceiving). He doesn't have a lot of lateral range, but in the past he has positioned himself well, and despite his lousy arm gets the outs that it seems he should. . . . I think Cardinals fans will be more than pleased with his effort afield." adds waterloo: "The Angel pitching staff has been *heavily* flyball oriented the past few years, so there just aren't as many groundballs for their infielders to field. So range factor will undervalue the Angel IFs just a bit ..."

rob haneberg makes a similar point over at the outstanding birdwatch blogplex, in a comment to the feb 8 post: "MGL's UZR and Tom Tippett's Diamond Mind ranking say Eckstein should be at least average in 2005. . . .Eck's range factor is below average but the Angels have had a flyball staff, something like 28th in the majors in GB/FB ratio in 2004, according to ESPN.com."

those are excellent points, from guys who are far stat-savvier than i. but i’m still concerned about eckstein’s range, and here’s why: in all four of his big-league seasons, eckstein has made fewer plays at shortstop than the guys backing him up at the position.

and it’s not even very close. over his career, eckstein has made 4.21 plays per 9 innings (vs a league average of 4.54 plays per 9). in that same four-year period, the angels’ other shortstops — playing behind the same flyball-oriented staff as eckstein — have made 4.81 plays per 9, a spread of more than half a play per game. here’s the data:

ANGELS SS, 2001-2004:
eckstn 4503 inn, 776 putouts 1332 assists = 4.21 plays per 9 innings
others* 1268 inn, 277 putouts 400 assists = 4.81 plays per 9 innings

* alfredo amezaga, chone figgins, shane halter, benji gil, wilson delgado, jose nieves, troy glaus

i know, i know — using range factor these days is sort of like using a compaq 286. ev’body’s moved on to vastly more powerful computational tools like uzr, pmr, defensive win shares, and so forth to measure defensive performance. the problem with those methods is that they are so often in conflict — a gold glover by uzr is too often doctor strangeglove according to pmr or dws — that you never know which one(s) you can trust. and they’re so new that it’s difficult to benchmark the numbers. range factor may be primitive, but at least it’s transparent; you don’t have to be a math ph.d. to understand how it works. i think as long as you work within its limitations, range factor can still tell a truth or two

one of the limitations is that you’re usually not comparing apples to apples — the fielding context (ballpark, pitching staff tendencies, teammates’ fielding range) varies from player to player at a given position, rendering comparisons suspect. but in this case we’re comparing a group of players who did their fielding within the same context — same team, same pitchers, same ballpark, same playing surface. and the comparison doesn’t favor eck —indeed, he has underperformed his understudies every single season of his career. the year-by-year breakdown:

2001 deficit: -0.5 pp9
eckstn 1051 inn, 178 po, 333 a, 4.37 pp9
others 387 inn, 87 po, 122 a, 4.87 pps

2002 deficit: -0.83 pp9
eckstn 1276 inn, 207 po, 397 a, 4.26 pp9
others 175 inn, 35 po, 64 a, 5.09 pp9

2003 deficit: -0.07 pp9
eckstn 985 inn, 193 po, 293 a, 4.44 pp9
others 445 inn, 95 po, 128 a, 4.51 pp9

2004 deficit: -1.19 pp9
eckstn 1191 inn, 198 po, 309 a, 3.83 pp9
others 262 inn, 60 po, 86 a, 5.02 pp9

that sure looks like a pattern . . . . true, the backups logged limited playing time in each season, so random chance might explain the disparity for any year taken in isolation. but over the four seasons the backups have amassed an entire season’s worth of innings — and their range factor over that "season" is .37 pp9 better than eckstein’s best season.

which leaves me a lot less persuaded that eckstein’s apparent lack of range can be explained by the angels’ flyball-oriented pitching staff. he may in fact be a more effective defender than his range factors suggest — but i’m still looking for evidence to support the notion.