BABIP and Player Styles
Often times, the analysis on this, and many sabermetrics inclined sites, rely heavily on batting average on balls in play (BABIP) to indicate luck in a hitters profile. Peter has done extensive work on BABIP and batted balls, but I wanted to take a different look at the concept. My focus was not on what components weighed heavily in BABIP, but rather what player types, if any, conveyed unnatural BABIPs. Generalizations about styles like three true outcome hitters and speedsters are usually made in the context of a conversation about BABIP, are those stereotypes true, or simply misconceived?
The only overall qualifications were: active post-1980 and 5,000+ plate appearances. Here are the player styles tested:
Three True Outcomes (Top 50 were tested)
Essentially 3TO types are the players who either walk, strikeout, or hit a homerun in a disproportional amount of plate appearances. Think Jim Thome, Mark McGwire, Pat Burrell, or even Jay Buhner. These players are bulkier, slower, defenseless, and possess what Bill James labeled "Old player skills", although they're not always old.
Speedsters (Top 50 were tested)
This category is exactly what you're thinking it consists of. These guys don't hit homeruns, but do steal bases, and generally make good usage of their speed in the field. Rickey Henderson, Vince Coleman, Kenny Lofton, and Trot Otis Nixon are grouped here.
The Non-Three True Outcomes (Top 50 were tested)
Exact opposite of 3TO. Juan Pierre, Placido Polanco, Lance Johnson, and Rey Sanchez don't tend to homer, walk, or strikeout.
Now, on to the results.
| playerID | BABIP | 3TO |
| abreubo01 | 0.3473 | 36.2345 |
| ramirma02 | 0.339 | 37.8276 |
| vaughmo01 | 0.3382 | 38.7207 |
| galaran01 | 0.3362 | 33.4829 |
| lankfra01 | 0.3298 | 39.2557 |
| edmonji01 | 0.3241 | 39.2959 |
| hernajo01 | 0.3237 | 38.4752 |
| leede02 | 0.3228 | 35.5438 |
| rodrial01 | 0.3222 | 35.0331 |
| thomeji01 | 0.3219 | 47.4191 |
| berkmla01 | 0.3219 | 36.6776 |
| jenkige01 | 0.3219 | 34.7553 |
| tartada01 | 0.3218 | 40.9589 |
| posadjo01 | 0.3197 | 37.2552 |
| salmoti01 | 0.3166 | 37.3491 |
| bagweje01 | 0.3165 | 36.1476 |
| sandere02 | 0.3118 | 36.89 |
| luzingr01 | 0.3112 | 35.2323 |
| daviser01 | 0.311 | 39.388 |
| mcgrifr01 | 0.3095 | 36.1777 |
| varitja01 | 0.3058 | 34.1298 |
| ortizda01 | 0.3042 | 36.5463 |
| thomafr04 | 0.3041 | 35.5867 |
| camermi01 | 0.3033 | 38.8463 |
| delgaca01 | 0.303 | 38.5138 |
| clarkto02 | 0.3023 | 38.6355 |
| sosasa01 | 0.301 | 38.9108 |
| gibsoki01 | 0.3009 | 34.0419 |
| giambja01 | 0.3008 | 37.4196 |
| burrepa01 | 0.2997 | 42.8545 |
| cansejo01 | 0.2989 | 40.7234 |
| jacksre01 | 0.2984 | 39.7702 |
| barfije01 | 0.2948 | 37.672 |
| sexsori01 | 0.293 | 39.3826 |
| buhneja01 | 0.2874 | 42.4007 |
| stairma01 | 0.2874 | 34.4495 |
| palmede01 | 0.2851 | 38.2828 |
| bondsba01 | 0.2845 | 38.5574 |
| cruzjo02 | 0.2823 | 37.0323 |
| strawda01 | 0.2817 | 39.5731 |
| glaustr01 | 0.2814 | 40.4281 |
| burnije01 | 0.2787 | 36.9863 |
| fieldce01 | 0.2786 | 39.1985 |
| gantro01 | 0.2786 | 34.2365 |
| jonesan01 | 0.2779 | 34.4299 |
| tettlmi01 | 0.2778 | 43.6932 |
| valenjo03 | 0.2717 | 34.7847 |
| johnsho01 | 0.2694 | 34.5776 |
| vaughgr01 | 0.2611 | 38.7165 |
| mcgwima01 | 0.2553 | 45.6576 |
3TO% BABIP R2: 0.232
3TO% BABIP standard deviation: 0.02
3T0% AVG BABIP: 0.302
---
| playerID | BABIP | SB |
| henderi01 | 0.3052 | 1406 |
| colemvi01 | 0.3145 | 752 |
| wilsowi02 | 0.3291 | 668 |
| loftoke01 | 0.3265 | 622 |
| nixonot01 | 0.3082 | 620 |
| smithoz01 | 0.2751 | 580 |
| butlebr01 | 0.3193 | 558 |
| lopesda01 | 0.281 | 557 |
| bondsba01 | 0.2845 | 514 |
| molitpa01 | 0.3259 | 504 |
| alomaro01 | 0.3215 | 474 |
| younger01 | 0.2938 | 465 |
| deshide01 | 0.3137 | 463 |
| saxst01 | 0.2989 | 444 |
| pierrju01 | 0.3165 | 429 |
| grissma02 | 0.2946 | 429 |
| biggicr01 | 0.3107 | 414 |
| knoblch01 | 0.3112 | 407 |
| samueju01 | 0.3135 | 396 |
| collida02 | 0.3069 | 395 |
| vizquom01 | 0.2941 | 385 |
| larkiba01 | 0.3065 | 379 |
| smithlo01 | 0.3256 | 370 |
| womacto01 | 0.3055 | 363 |
| damonjo01 | 0.3074 | 362 |
| mcgeewi01 | 0.341 | 352 |
| daviser01 | 0.311 | 349 |
| whitede03 | 0.3051 | 346 |
| sandbry01 | 0.3043 | 344 |
| castilu01 | 0.3313 | 342 |
| otisam01 | 0.2946 | 341 |
| wilsomo01 | 0.3181 | 327 |
| johnsla03 | 0.3069 | 327 |
| polonlu01 | 0.3215 | 321 |
| finlest01 | 0.2852 | 320 |
| gwynnto01 | 0.3414 | 319 |
| abreubo01 | 0.3473 | 318 |
| cruzjo01 | 0.3066 | 317 |
| suzukic01 | 0.3543 | 315 |
| anderbr01 | 0.2818 | 315 |
| sandere02 | 0.3118 | 304 |
| rolliji01 | 0.2982 | 295 |
| camermi01 | 0.3033 | 289 |
| dykstle01 | 0.3041 | 285 |
| baylodo01 | 0.2602 | 285 |
| gibsoki01 | 0.3009 | 284 |
| rodrial01 | 0.3222 | 283 |
| francju01 | 0.3331 | 281 |
| renteed01 | 0.3198 | 280 |
| mosebll01 | 0.2908 | 280 |
SPD BABIP R2: 0.0017
SPD BABIP standard deviation: 0.018
SPD AVG BABIP: 0.309
---
| playerID | BABIP | 3TO |
| foliti01 | 0.2623 | 10.7589 |
| bucknbi01 | 0.2866 | 10.7851 |
| guilloz01 | 0.2797 | 11.127 |
| pierrju01 | 0.3165 | 11.4843 |
| rayjo01 | 0.298 | 13.1039 |
| polanpl01 | 0.3146 | 13.3346 |
| gwynnto01 | 0.3414 | 13.3405 |
| johnsla03 | 0.3069 | 13.3565 |
| gantnji01 | 0.2897 | 13.9455 |
| sanchre01 | 0.2986 | 14.5934 |
| perezne01 | 0.2835 | 14.7477 |
| jeffegr01 | 0.2878 | 15.59 |
| saxst01 | 0.2989 | 15.7999 |
| smithoz01 | 0.2751 | 15.9883 |
| ramirra01 | 0.284 | 16.1335 |
| mattido01 | 0.3002 | 16.2688 |
| suzukic01 | 0.3543 | 16.5736 |
| burleri01 | 0.29 | 16.8116 |
| kendaja01 | 0.31 | 16.8907 |
| oberkke01 | 0.2924 | 16.9736 |
| cabreor01 | 0.2849 | 16.9852 |
| baergca01 | 0.3024 | 17.1239 |
| boonebo01 | 0.2622 | 17.1871 |
| reynoha01 | 0.2769 | 17.2492 |
| younger01 | 0.2938 | 17.3756 |
| fernato01 | 0.3076 | 17.9693 |
| polonlu01 | 0.3215 | 18.0262 |
| lansfca01 | 0.3021 | 18.1228 |
| surhobj01 | 0.2915 | 18.4505 |
| greenmi01 | 0.3046 | 18.4776 |
| fletcsc01 | 0.2844 | 18.5678 |
| simmote01 | 0.2837 | 18.5756 |
| womacto01 | 0.3055 | 18.6234 |
| vizquom01 | 0.2941 | 18.648 |
| grudzma01 | 0.3227 | 18.6728 |
| templga01 | 0.3067 | 18.8566 |
| reedjo01 | 0.2899 | 18.9184 |
| loretma01 | 0.3169 | 18.9743 |
| knighra01 | 0.2903 | 19.0928 |
| whitefr01 | 0.2743 | 19.2087 |
| garcino01 | 0.3117 | 19.3033 |
| beniqju01 | 0.2948 | 19.3364 |
| trillma01 | 0.2889 | 19.3524 |
| wilsowi02 | 0.3291 | 19.5081 |
| vizcajo01 | 0.3046 | 19.6696 |
| brettge01 | 0.3069 | 20.0121 |
| penato01 | 0.2835 | 20.0999 |
| hamilda02 | 0.3155 | 20.2814 |
| kotsama01 | 0.3 | 20.3067 |
| gracema01 | 0.3092 | 20.3818 |
N3T BABIP R2: 0.0173
N3T BABIP standard deviation: 0.017
N3T AVG BABIP: 0.299
Through all of that, there's not much of a trend here. Generally, hitter BABIP falls between .290-.310, as each of the averages, and most of the ranges do. Each style has its own outliers, like Ichiro, Bobby Abreu, Greg Vaughn, and Don Baylor, but there's not enough here for us to estimate who or what makes for those kind of breaks. This hardly the definitive post on BABIP and player styles, and you could make the case that speed scores should've been used rather than stolen bases, however I'm just attempting to provide a general idea, and at this point it seems to be that there is no BABIP specialization.
20 comments
|
0 recs |
Do you like this story?
Comments
Interesting work. I think an alternative way to get at this would be to regress all players 3TO% versus BABIP and see if there is any trend.
for all players in 2008 with >100 PAs I find no realationship between 3TO% and BABIP. I tried an artifical ‘speed index’ which was sb/(bb+1b) and found that babip was correlated with sb/(bb+1b) for players with >100 PAs last year. The relationship was positive with r^2=.011 so explaining a small amount of variation but significant (p=.025). The slope of the line was ~.06 so a player with an sb/(bb+1b) of .1 should have a higher babip by .006 compared with a player with an sb/(bb+1b) of 0.
Also: the "speed" category really does need work.
SBs are hardly the best way to say a player has speed.
by R.J. Anderson on Nov 23, 2008 1:16 PM EST up reply actions
Did you really mean Trot Nixon?
I am guessing you are really meaning Otis? Not to take anything away from your analysis though…very interesting work…great read!
"I don't want to play golf. When I hit a ball, I want someone else to go chase it." -Rogers Hornsby
Is it true that the lowest and highest of the list can be written off as extreme cases
or is there some truth in the fact that those who have a high 3TO rate will hit the ball hardest or hit the most fly balls? Many of the guys at the bottom of this list had a high percentage of their hits be HRs, and hit a lot of flies, and were also very slow. There are not a lot of 50 home run hitters in the game, especially today, so maybe we can call these extreme examples and a dying breed because of the steroid testing… and so then, it would be prudent to see whether the top of the list are also extreme examples, and my gut feeling tells me that they are not, and I do not have the time at the moment to see if this is true…
by Daniel Berlyn on Nov 23, 2008 1:59 PM EST up reply actions
Interesting analysis, confirming what we think we know
One thing I’m curious about is the effects of a great bunter on BABIP. Alcides Escobar carried a .380 BABIP at AA this year, and he obviously can’t keep that up, but he was something like 15/30 on bunt hit tries this year. I wonder if he’ll be a guy that can keep a higher BABIP if he’s able to have that kind of base-hit bunting sucess at higher levels.
And neck size to baby eating ratio.
i would argue
that bunts and bunt attempts should not be considered in BABIP, since BABIP is used to measure luck and bunting is definitely a skill.
god, i love baseball. -roy hobbs
Hitters' BABIP is NOT LUCK.
There are many ways to have above-average BABIP, and the best one we know of so far is to hit lots of line drives. Some guys, eg Michael Young, do that. His career BABIP is .339 and projects to stay that way as long as he keeps hitting LDs.
The purpose of this study wasn’t to show “BABIP=luck,” which is clearly wrong, it was to show that these 3 particular variables— speed, TTO, and (1-TTO)— aren’t particularly relevant to BABIP. It’s like a differential diagnosis: we’ve knocked these out as major factors, which means something else is behind it. I believe, personally, that the missing variable will turn out to be batted ball velocity (speed of the ball off the bat) once that starts getting measured in a few years’ time.
Your 2008 Athletics: It's Nothing Personal.
sometimes it is, sometimes it isn't
Just looking at a guys BABIP and calling it lucky if it is high is wrong, of course. But if you just look at a players LD rate and his GB and FB rates, a guy who bunts for hits 10 times in a season with a 50% BUH rate (which get recorded as ground balls and infield hits by fangraphs/B-R) will seem to have a very lucky GB BABIP when in reality he has a specific skill that another player may not have.
If you are trying to predict what he’ll do next year, you’d expect his GB BABIP to regress, but as long as he maintains his bunting skills that is a poor expectation.
Anyway, while we’re on this topic, I agree with Azr below. A better study would look at power hitters and see if they have higher or lower than expected BABIP on GB/FB/LD etc.
god, i love baseball. -roy hobbs
Without accounting for the batted ball compilation
I’m not sure what this study was attempting to do. A better angle would seem to be measure the difference between BABIP and eBABIP for each group.
The biggest problem with this study is selective sampling issues.
It’s like the old analogy of concluding that size doesn’t matter for offensive lineman, because you can’t find a correlation between weight and talent in the NFL. Of course bigger is better! But if a smallish guy DOES make the NFL, he just has tons of other skills that make up for lack of weight.
I’m no statistician, but I believe a multi-variable regression analysis is what you want here, which will effectively hold all other variables constant while you test the effects of individual variables.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by 


















