Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Transaction Oracle > Discussion
Transaction Oracle
— A Timely Look at Transactions as They Happen

Tuesday, November 27, 2007

Fun with Translation: BABIP

As some of you know, trying to translate minor league performance is one of my passions in baseball analysis, and not just for the motive of improving my projections (I’ve been working with them since the 1990s).  I’ve tinkered with a lot of the original MLE formula of James (as have Davenport and Smith and others) and I’m now working one of my pet projects - working on BABIP and SLGIP for minor league hitters.  We had a bit of a discussion about this in one of the Lidge trade threads, when talking about Mike Costanzo.  I’ve theorized that one of the troubles with translating high strikeout batters is only indirectly a strikeout issue, but more likely a BIP issue, that high strikeout totals and success means a player generally has an unsustainable BABIP.  So I had extra time this holiday weekend, being that it rained non-stop and I ducked family and I sure as hell wasn’t going shopping anywhere.  Using data going back to 1992, thanks to Sean now having the data up handily, I found a few interesting things that look like they’ll improve my minor league translations.

First, it really looks like figuring out the number of balls hit into the field of play and the rate at which good things happen to those balls hit into player.  Instead of a conversion rate for hits and the like, I’m getting more predictive results by finding the conversion rates for BABIP and finding the conversion rates for % of balls into play with a certain result.  Pretty typically, a batter loses between 4 and 6% of their BABIP going from AAA to the majors, 3-5% from AA to AAA and less for lower level promotions.  The change is small, but looking at my group of matched sets of low and high strikeout batters, it’s enough to eliminate the difference in projectability.

Just to use Mike Costanzo as the base (with 157 strikeouts), here are my MLEs for him with varying numbers of strikeouts:

200:  214/282/383
180:  218/285/392
157:  223/290/400
120:  229/296/413
100:  233/299/417
80:  235/301/420
60:  238/304/424
40:  240/306/432

OK, hardly a huge change, but we’re at the point with this particular stuff at which most of our changes to various concepts are going to be minor.  Not sure if someone else has taken a look at this before and it would hardly be the first time I replicated work already done (Chone and I both spent a helluva lot of time in September making a rough translation of minor league defensive numbers to a rudimentary zone rating without knowing each other were doing it), but I thought I’d put it up just in case someone was interested.

My next project this winter?  Looking back on the last 15 years not an eye on minor league—>major league numbers but an estimate of minor league projection—->major league numbers.  I’m kinda wondering if a Marcel-type look at play in the minors instead of seasonal lines might wring out a little bit of noise.

Dan Szymborski Posted: November 27, 2007 at 08:15 AM | 33 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. AROM Posted: November 27, 2007 at 03:10 PM (#2625952)
Cool stuff. I can't remember if I do this in my mle's or just in projections. I think I handle all that in the projection and regress K rate and babip seperately, but if you do it at the projection stage you've got multi-year data, so you don't have to regress as much.

We should compare notes on the minor league fielding ratings. I think I had most of AA and AAA done but haven't looked at it in almost 2 months. Maybe we can combine our stuff and publish it here.
   2. snapper (history's 42nd greatest monster) Posted: November 27, 2007 at 05:09 PM (#2626090)
I’ve theorized that one of the troubles with translating high strikeout batters is only indirectly a strikeout issue, but more likely a BIP issue, that high strikeout totals and success means a player generally has an unsustainable BABIP.

Interesting Dan.

So, are you saying that it is a selection bias, i.e. the high strike hitters that get promoted are the lucky ones who have unusually high BABIP?

But shouldn't the low strikeout hitters that get promoted also be selected in the same way?

Thanks for all your work!
   3. Dan Szymborski Posted: November 27, 2007 at 05:25 PM (#2626106)
So, are you saying that it is a selection bias, i.e. the high strike hitters that get promoted are the lucky ones who have unusually high BABIP?


That's probably true to some extent, but all players that are having better-than-true-talent seasons are more likely to be promoted (and disappointing). There are selection biases that run both ways in MLEs.

It's just that overall (and this is true for the majors and BABIP as well), that the rate at which good things happen when balls are put in play has more variability than the rate at which the ball is actually put in play.

A player that relies on putting a few balls into play and having a lot of them not be outs is inherently going to be have wider performance swings - take a look at Jack Cust and Russ Branyan's careers. They're not getting high BABIPs through their legs, like Chone Figgins.

Now, I'm not saying this effect is huge or a reason to go ballistic over strikeout it's just something to keep in mind when a guy like Justin Ruggiano has a huge BABIP year.
   4. Mike Green Posted: November 27, 2007 at 05:30 PM (#2626113)
It seems to me that when looking at the meaning of minor league BABIP, some studies controlling for age would be the first step. The Midwest League results of Jay Bruce, Cameron Maybin and Travis Snider raise this issue squarely. It would be interesting to compare the careers of players in the Midwest League with "unsustainable" BABIPs at age 19, 20, 21 and 22.
   5. Dan Szymborski Posted: November 27, 2007 at 05:41 PM (#2626124)
I did general groupings for age and quality as well and found no real difference in BABIP dropoff.

One problem we have is that our estimates of translating lower leagues are based off indirect results - players from low and mid-A rarely go straight to the majors and so we have to estimate how to do it by seeing how they do in AA or AAA and working from there.

That's part of the reason I want to look at some of these data using a projection instead of the actual seasonal data so we can hopefully weed out some of the changes in player ability from year-to-year that aren't related to league change but to normal development.
   6. snapper (history's 42nd greatest monster) Posted: November 27, 2007 at 05:49 PM (#2626130)
A player that relies on putting a few balls into play and having a lot of them not be outs is inherently going to be have wider performance swings -

OK, so you're saying it's more of a sample size issue. If you have one palayer with 600 PA, with 30 BB, and 30 K, and another with 600 PA, 150 BB and 150 K, the BABIP for your 540 BIP hitter is likely to be closer to his true talent BABIP, than for the 300 BIP hitter, just due to sample size.
   7. Kyle S Posted: November 27, 2007 at 06:10 PM (#2626160)
That's part of the reason I want to look at some of these data using a projection instead of the actual seasonal data so we can hopefully weed out some of the changes in player ability from year-to-year that aren't related to league change but to normal development.

I think this should be done (if it hasn't already) with translating Japanese performance as well. My translations using "matched seasons" worked pretty well, but a better way to create them would be to do a 3/2/1 (or whatever) projection into Japanese stats (accounting for aging, etc), then pro-rate the stats for actual American PAs, and compare that way. Is that how Clay does it already?
   8. Mike Emeigh Posted: November 27, 2007 at 07:28 PM (#2626259)
the high strike hitters that get promoted are the lucky ones who have unusually high BABIP?


Suppose you have a guy who projects to walk 100 times and fan 180 times in 600 PA. That leaves him with 320 non-BB, non-K PA. Give him 30 HR, you're down to 290 BIP. If he bats .300 on those BIP, that's 87 non-HR hits, which probably have about 30 doubles and maybe a triple. That's a .234/.362/.478 line, which isn't going to get anyone promoted.

Bump it to .350 on BIP. Now you have 102 non-HR hits, of which 35 are probably doubles and still 1 triple. Now he's at .264/.387/.518; better, but still not a lock to be promoted.

At .400, with 40 doubles and a triple, you have 116 non-HR hits. He's now hitting .292/.410/.556, and he WILL get promoted at that point.

Is he "lucky" to be batting .400 on BIP? Perhaps, but more likely his true talent isn't far off that. High-strikeout hitters simply have to do well on contact in order to survive in the majors; if they don't, the strikeouts do them in.

Segueing to Hitter Evaluation 101:

The two most important things that you need to know about a minor league hitter are:
1. how frequently he makes contact;
2. what he does when he makes contact.

The worse he is at #1, the better he *must* do at #2.

-- MWE
   9. Dan Szymborski Posted: November 27, 2007 at 07:38 PM (#2626267)
Woo, I'm glad Emeigh was in this thread since I know he knows the point I'm trying to make and he expresses it a million times better than me! I can use a translator from Dan to English once in a while.
   10. Walt Davis Posted: November 27, 2007 at 07:44 PM (#2626272)
that high strikeout totals and success means a player generally has an unsustainable BABIP.

Welcome to the party. :-) But they also need hellacious power totals and I don't think SLGBIP is the way to get at that because you're missing the HRs and you want SLG on-contact. I don't really see any inherent advantage in using BABIP over BA on-contact either, but it could be the former is substantially more variable. But with SLGBIP, I think very little of the major-league success of these players has to do with how many doubles they hit (and they're rarely major triples hitters).

For example, Howard's mid-20s in doubles, Sosa had a few nice years in the 30s but was usually in the low 20s, Thome has usually been in the 25-30 range, McGwire never over 28. At the more human end, Rob Deer never more than 24, Mark Bellhorn was a pretty good doubles hitter, Jose Hernandez not so much.
   11. snapper (history's 42nd greatest monster) Posted: November 27, 2007 at 07:53 PM (#2626283)
Is he "lucky" to be batting .400 on BIP? Perhaps, but more likely his true talent isn't far off that. High-strikeout hitters simply have to do well on contact in order to survive in the majors; if they don't, the strikeouts do them in.

I understand why hig K hitters need the high BABIP. I'm just struggling with why it will decline more going to the majors.
   12. Dan Szymborski Posted: November 27, 2007 at 07:54 PM (#2626285)
Oh, when I look at SLGIP, I'm looking at homers, too. In fact, I found that SLGIP doesn't seem to add any translation utility over and beyond what BABIP and simply measuring the change in hit distribution do.
   13. AROM Posted: November 27, 2007 at 07:55 PM (#2626286)
Doubles:

There is very little variablility in the double rates of most players. You have a few outliers, the super-aggressive baserunners like Biggio and the painfully slow like Frank Thomas, but most players hit about 25-35 per year, regardless of whether they are little guys or power hitting monsters. What variation you see (i.e. Bellhorn vs Jose Hernandez) is often the result of ballparks. Bellhorn hit 37 per 500 AB in Fenway, but only 24 per 500 in other parks.
   14. Dan Szymborski Posted: November 27, 2007 at 07:55 PM (#2626287)
I understand why hig K hitters need the high BABIP. I'm just struggling with why it will decline more going to the majors.

Well, that's the question! We can measure what, but why is more hypothetical.
   15. KJOK Posted: November 27, 2007 at 07:59 PM (#2626291)
Pretty typically, a batter loses between 4 and 6% of their BABIP going from AAA to the majors,..


Are you finding that BABIP for the majors overall is 4-6% lower than in AAA? That would have repercussions for pitchers going from AAA to MLB that would be POSITIVE for them...this is somewhat what I'm finding for Japanese pitchers..
   16. AROM Posted: November 27, 2007 at 08:06 PM (#2626298)
Are you finding that BABIP for the majors overall is 4-6% lower than in AAA? That would have repercussions for pitchers going from AAA to MLB that would be POSITIVE for them...


Not really, since the quality of play is not equal. Batters get fewer hits moving up, and pitchers give up more. I think for the PCL pacific conference there's so much offense that pitchers from there do better in the majors on BABIP, but that's the exception - and batters from that conference lose a lot more than the typical 4-6%.
   17. dr. bleachers Posted: November 27, 2007 at 08:15 PM (#2626311)
I thought BABIP differences moving up was mostly defense related, along with pitchers in the majors being better at controlling BIP. Maybe I'm not really following you guys.
   18. KJOK Posted: November 27, 2007 at 08:31 PM (#2626323)
I thought BABIP differences moving up was mostly defense related, along with pitchers in the majors being better at controlling BIP. Maybe I'm not really following you guys.


That's what I see, but I think AROM is disputing this? There are multiple factors that impact batters/pitchers, and they impact them in DIFFERENT directions for some factors:

1. Quality of play - impacts both batters and pitchers negatively moving to MLB.
2. BABIP - fielding is better in MLB, so impacts batters negatively, but pitchers POSITIVELY moving to MLB?
3. Ballparks - MLB parks are on average more pitcher friendly, so impacts batters negatively, but pitchers POSITIVELY.
4. League scoring - For high scoring league like PCL, impacts batters negatively, but pitchers POSITIVELY.

Adding it all up, I think you'll find that batters take a bigger 'hit' moving from AAA to MLB than pitchers do, especially from PCL I would think...
   19. Kyle S Posted: November 27, 2007 at 08:54 PM (#2626342)
Well, the average pitcher in AAA probably has a lot less BABIP prevention talent than the average pitcher in MLB. Even figuring that the better pitchers tend to be selected to advance, they are still likely to be no better than average. The hitters they face in MLB will be much better at hitting the ball hard than the hitters they used to face. I think these two factors would swamp the others (quality of fielding, ballpark).

I don't think there's a separate "league" adjustment apart from the park adjustment - scoring is high in the PCL relative to the IL not because the league is better but because it has smaller parks/more parks at high altitude on average.
   20. AROM Posted: November 27, 2007 at 08:58 PM (#2626351)
I thought BABIP differences moving up was mostly defense related


The only thing I'd dispute is the word "mostly". Its a factor but I don't think its bigger than the fact that major leaguers are tougher to get out than AAA hitters.

KJOK, That list seems good to me. I think its hitters taking a bigger hit from the PCL, and pitchers taking a bigger hit moving up from the International League. Overall, I'm not sure if pitchers or hitters take a bigger hit but its pretty close.
   21. dr. bleachers Posted: November 27, 2007 at 09:09 PM (#2626367)
The only thing I'd dispute is the word "mostly". Its a factor but I don't think its bigger than the fact that major leaguers are tougher to get out than AAA hitters.

I've always assumed that pitcher and hitter BIP talent increases through the levels in such a way to cancel each other out somewhat. The park and league factors vary quite a bit (though maybe the minors have a prevalence of hitters' leagues like the PCL, I'm not well versed) and just are what they are. They don't really speak to true talent obviously. So what you are left with is defense that affects BABIP on a league-wide level, along with the pitchers that hang around MLB forever having BIP talent. Though you could say the same about hitters too I guess.

When talking about the individual hitter or pitcher, you can assume all these things affect them in the way you would think in a MLE sense.
   22. Mike Emeigh Posted: November 27, 2007 at 10:26 PM (#2626477)
I thought BABIP differences moving up was mostly defense related, along with pitchers in the majors being better at controlling BIP. Maybe I'm not really following you guys.


hBIP is primarily a function of how hard the batter hits the ball, less importantly where. It's something like 60-40 batter/defense (where defense=pitching+fielding) on the part we can measure, with a fairly large variance.

I've always assumed that pitcher and hitter BIP talent increases through the levels in such a way to cancel each other out somewhat.


They don't, quite, for a couple of reasons:

1. Hitters have to play defense;
2. There are fewer hitters who can hit at the major-league level than there are pitchers who can pitch at the major-league level.

-- MWE
   23. snapper (history's 42nd greatest monster) Posted: November 27, 2007 at 10:26 PM (#2626478)
Well, that's the question! We can measure what, but why is more hypothetical.

OK. Clear now. I thought I was being dense.
   24. dr. bleachers Posted: November 27, 2007 at 10:37 PM (#2626486)
Ah, good stuff. Thanks, Mike.
   25. Mike Emeigh Posted: November 27, 2007 at 11:01 PM (#2626508)
There are fewer hitters who can hit at the major-league level than there are pitchers who can pitch at the major-league level.


When I say this, by the way, what I am talking about is that there are more pitchers who can pitch in some role at the major league level. I think there are fewer prospects who can pitch in a front-end role as there are hitters who will hit enough to be a regular, but there are more pitchers who can fill out a staff than there are hitters who will fill out a bench. In large part, I think, that's why teams have moved toward having more pitchers on the roster; it's relatively easier to find a pitcher who can fill a low-leverage role than it is to find a hitter who can thrive in part-time duty.

-- MWE
   26. kwarren Posted: November 27, 2007 at 11:01 PM (#2626510)
AROM - I know that this is off topic, but do you have any idea when pitching projections will be available.

I just discovered your hitting projections yesterday and am finding them useful, and apparently of high quality. Thanks for doing them.
   27. Russ Posted: November 27, 2007 at 11:18 PM (#2626527)
it's relatively easier to find a pitcher who can fill a low-leverage role than it is to find a hitter who can thrive in part-time duty.


That's an interesting theory Mike... I think that this also will be self-reinforcing, in that straight platoons are far less common nowadays, probably because they are negated by bigger bullpens. The more fungible pitchers become during the game, the less having a straight platoon can help you. It would be interesting to see a team try to run with a much smaller pen with more bats to try and grab back the platoon advantage. I think that's tough because pitchers do much better in a short situation than hitters (probably because short for pitchers is three BF, short for hitters is 3 times less -- one AB).
   28. greenback calls it soccer Posted: November 28, 2007 at 01:01 AM (#2626595)
I'm looking forward to the day when the pitch fx stuff shows up in minor league ballparks.
   29. AROM Posted: November 28, 2007 at 01:20 AM (#2626607)
AROM - I know that this is off topic, but do you have any idea when pitching projections will be available.


They are ready, and thanks.

Here's link.
   30. Los Angeles Waterloo of Black Hawk Posted: November 28, 2007 at 02:11 AM (#2626640)
First, it really looks like figuring out the number of balls hit into the field of play and the rate at which good things happen to those balls hit into player.

Who wrote this sentence, Chris Dial?

***

Why do we care about BABIP for hitters? If, as Mike says, the things you want to know about a hitter are (1) how often he makes contact and (2) how well he does when he does make contact, don't we want to include HR in that?
   31. Walt Davis Posted: November 28, 2007 at 07:18 AM (#2626820)
Why do we care about BABIP for hitters? If, as Mike says, the things you want to know about a hitter are (1) how often he makes contact and (2) how well he does when he does make contact, don't we want to include HR in that?

Moments like this bring a little tear of joy to my eye.

And make me realize nobody reads my posts. :-)
   32. Russ Posted: November 28, 2007 at 08:32 PM (#2627262)
One of the things that kills people in projections is that it is less likely that players with poor projections will get to live up to them. So, I would guess, that this would have a grave impact on the analyses and how well you would be able to predict people outside of the middle range. Players with poor projections would be unlikely to be able to obtain enough PT to nail down that bad projection, so the ones that are left are likely to have overperformed their projection. Similarly, players with very good histories (especially where lots of resources have been invested) are probably more likely to get more chances when they're *underperforming* projections.

That's why, at some level, it would probably be better to mix major and minor league numbers when evaluating the projections (which I think is what Dan is suggesting at the end above) and what AROM sort of suggests in his link.
   33. Mike Emeigh Posted: November 29, 2007 at 12:30 AM (#2627545)
If, as Mike says, the things you want to know about a hitter are (1) how often he makes contact and (2) how well he does when he does make contact, don't we want to include HR in that?


You want to look at in-play ISO first, because home runs are often a product of the environment. A guy who hits a lot of fly balls may wind up with 35 HRs in the PCL, or in a ballpark like Asheville's McCormick Field, without necessarily having "real" power.

-- MWE

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Francis
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.4980 seconds
47 querie(s) executed