Some Curious Things About BABIP

[RchW]

Originally Posted by Left-handed Badger

What


How was in the hands of the pitcher? Didnt the 2 just finish pointing out that those great plays were the difference in the game thereby undermining that?

A home run is strictly pitcher controlled. What led up to it was as described.

[lynchjm24]

Originally Posted by RchW

A home run is strictly pitcher controlled. What led up to it was as described.

Ah, I didn't get your point either. Richmond gave up a walk-off HR to Harris in the 12th to end it.

I was into it because I like when teams run out of pitchers like Toronto pretty much had and they used this coming Tuesday's starter in relief. Although Cito said they had Hayhurst available, he didn't use him.

[knockahoma]

knockahoma: Of course, I invite someone else to research that. Makes me do less work!

Lynch: Can you give me a 10 paragraph summary on how I misunderstood this too?

How about if I jam it all into half that many?

If you look over the past few threads I've contributed to, I'm constantly bringing stats and links. I've researched Belanger and various BABIP views, The height and weight of baseball players in the 70s vs the 2000s, Messersmith, BABIP league averages from 1954 to present, Bill James' articles on fielding, Tango Tiger's site on BABIP, Niekro, I'd just run Messersmith through about 10 different simulations with 3 different teams, I've actually gone through and analyzed on another thread how TOT/YR applied to team trades. I've examined many defensive players, such as Cedeno, Belanger, Harrah through TOT/YR. I spent a good half hour taking your thought on UZR to the interent, but couldn't find anything on 74, which is the target year in discussion. That's all in the past few days.

I had just finished three posts, researching Hampton's numbers in Denver, Messersmith's numbers in OOTP and in real life. And I had just said that I was going to do a comprehensive search on pitchers traded in mid-season to examine BABIP.

Then, I wrote,

Of course, I invite someone else to research that. Makes me do less work!

It was the equivalent of "Whew!" after more than two hours of being on the internet, hunting down the answer to questions asked. I was also hoping someone could actually send me to a site where such a study had been done on BABIP.

If you can read through my posts and think I don't relish doing my own research, then you haven't read through my posts. I love research, though I usually spend my time doing it for the show, which leads me to philosophy/religion, history and science, UFOs and on-line dating. Unfortunately, not so much baseball.

[knockahoma]

We don't need a cautionary tale.

Maybe you don't, Lynch. But I can go back to the last two BABIP threads, point you to other threads on the internet and show you people trying to shut down the conversation. I can't tell you the names, but if you ask me to, I'll go the the previous threads and bring back at least 5 to 10 posts saying, basically, the conversation on BABIP is over! Doctrine! If you need me to get those, I will. Maybe you'll see what I was thinking, then.

The rest of your post gets back on to BABIP itself and that interests me. I'll break down your post for comment:

If you want to put together evidence and try to draw conclusions from it... then do that. Don't randomly punch around baseball-reference picking an example here and an example there while ignoring hundreds of examples who don't meet the criteria for which you are looking.

Geez, man, come on! I've just begun. Even Bill James and Tango must have had at least 5 days to gather all their info and write their books!

It's pretty simple: There is a ton of evidence that a lot of what people used to consider pitching skill is actually defense. It's so intuitive that it's hard to believe how hard people will argue against it.

I agree with that. Now, before you slam me, stop and think about something. Who started the Mark Belanger thread? Who argued intensely that Belanger made a serious contribution to his team, despite a .225 BA?

Now, I just read Vogon and once again, his greater knowledge of the new research has saved the day for me. I also like what Tango wrote. But, there will come a day when we'll better see BABIP control in a single season.

It will come with MPH on balls in play, location of hits, etc. And that will help us to see what a fielder sees behind this pitcher, or that. What will it tell us when one pitcher's balls in play are on average 5 MPH slower than another's?

Just as a quick aside, I was thinking about line drives while we were walking the border collie last night. As a shortstop, a one-hopper was basically a line drive to me-- and almost always, the TOUGHER kind. The ball just happened to hit the ground before it got to the SS, just as a line drive does in CF. But in a pitchers Line drive rate, is such a one hopper to shortstop considered a line drive? Or is it a grounder? (don't answer that! I'll do my own research!

Makes it sound like such a less well hit ball on paper. In fact, the very first play I had in College at SS was a smokin' one hopper in that dreaded "in-between zone". It was recorded as a ground out to SS. But, it was hit much harder than that. And I don't mind telling you, after the play, I kicked around some dirt as the bench cheered, while I tried to get my heart from my throat back down to my chest. It was more fear of failure than of the ball.

Anyway, here's what Vogon posted:

Here you go: http://www.insidethebook.com/ee/inde..._dips_numbers/

You can get a pitcher's "true" BABIP by adding 3700 league average balls in play to his numbers.

Let's say we're looking at Greg Maddux in 1992. He had a .257 BABIP on 756 BIP. 3700/(756+3700)=.83, so we regress his BABIP 83% of the way back to the mean. Atlanta's BABIP that year was .287, so we'll use that. 83%*.287 + 17%*.257 = .282.

If all we know about Maddux is what he did in 1992, then we would assume a "true" BABIP of .282. His actual .257 BABIP was a combination of 17% skill and 83% fielding/ballpark/luck/etc. as best as we can tell.

We, of course, know a lot more about Maddux than just 1992. For his career he had 15,285 BIP. Regression equals 3700/(15285+3700) = .195, so we regress his career BABIP less than 20%. His actual BABIP was .286 versus a league average of .295. .286*80.5% + .295*19.5% = .288. That's our best estimate of his true BABIP for his whole career.

And here's what Tango said about BABIP, summed up, bold mine:

Tom Tango recently began a stats wiki page, which includes an entry for BABIP. In that entry he states “for a pitcher, seasonal BABIP is a largely unreliable measure of his skill. In order for this metric to do a good job in measuring his skill, you need several seasons worth of data. This leads to the myth that a pitcher's skill in hit prevention is mostly the product of luck: we can't see year-to-year consistency in the metric as the noise overwhelms the signal. But, as we increase the number of years, the signal can finally match the noise.”[/


BABIP - Sabermetrics

When I first started reading about BABIP on this site, the implications of that stat were relayed to me in the language of Voros McCracken. Since then, guys like Tango, Tippet, etc, have changed that language. I very much like how they're describing it.

I guess, in order to see how close you and I are, I'd ask you to comment on Vogon and Tango's language.

[knockahoma]

double post

[knockahoma]

And I guess since I'm back at first base again, I'll share again my musings.

Vogon's link has set some numbers on pitcher contribution to bABIP. I look at them and wonder,

" Do those numbers translate well with pitchers traded in mid-season?"

Before this all began, I was really interested in digging into that. It mattered with OOTP. I plan on going through several 70s seasons, posting each traded pitcher and looking at his BABIP.

Let's say that Messersmith's BABIP goes from .246 in LA to .256 in Atlanta. What is an average swing on BABIP with pitchers traded in mid-season, or even in the next season? Is it .15, or .50, or 1.00?

If we can emulate what history shows (regardless of how much is pitcher, defense, luck), if we can figure the parameters on how a BABIP will rise or fall in such instances, can that be anything but helpful to OOTP modeling?

So, let's say that 90% of pitchers traded see a BABIP swing between -.25 and a +.25 rate (for example, Messersmith goes from .236 to .260 after his trade).

If we see in OOTP all these pitcher BABIP rates from -.75 to .75, (for example, Messersmith goes from .236 to .302). If 30%, 40%, or 50% of pitchers experience that kind of swing, we know the model isn't working.

I see way too many pitchers in OOTP fluctuating from .235-.325. In my preliminary investigation, I do NOT see real-life fluctuations that wide-- ESPECIALLY in the 70s. Even if we attribute ALL of BABIP to defense and luck, in real life, they didn't make pitcher BABIPs fluctuate THAT much.

If we can then manipulate the BABIP to perform in the real life parameters, this can only help OOTP to be "more realistic", whether in historical or fictional.

That's my thinking, anyway.

[lynchjm24]

Originally Posted by knockahoma

I guess, in order to see how close you and I are, I'd ask you to comment on Vogon and Tango's language.

The difference between us is that I'm consistent and you jump all over the place.

I'm really not in search of a holy grail that tells me that the pitcher's effect on hits on balls in play is 17% or 19% or 51%. I think that for every pitcher, in every stadium in every era with every defense that is a moving target.

To be honest you are wasting your time trying to get historical leagues to meet your expectations. The game can't properly import defensive ratings based of what's in the database.

Based on Ron's work on the FOBL it seems to me that defense is pretty well modeled in OOTP. Without adding a BABIP fudge factor for historical pitchers there is no way for the numbers to line up to history.

Could the OOTP model be tweaked to provide better results? Probably. But you need to approach it like Ron does with huge amounts of data and systematic studies, not picking a handful of guys from history.

I've got no real argument for Tango's point on regressing Maddux so little. The problem is that <1% of pitchers ever get that many balls in play.

[knockahoma]

Lynch,

I'm sorry to see it that way. Again, you seem like a pretty decent guy. For the record, I think Ron is too. Discussions get heated. And if I thought otherwise, I'd tell ya, and you wouldn't need a follow-up exegesis to understand how I felt about you.

But, basically, you're telling me to shut down the conversation, right? I'm wasting my time. So, let's compromise.

You don't have to waste your time on me and I'll keep on my happy way. A long time ago I found a happy secret. Good discussions lead to learning things along the way, which leads to thinking for oneself. That's really my attraction to most discussions. You can argue about WWII, or Predeterminism, wind up disagreeing, and yet, look how much was learned in the process! You learned about Chamberlain; you learned about behavioral studies, and how the explosion of language in children affects Free Will arguments.

So, no. It's not a waste of time for me. Because I'm not trying to win someone to my side. I'm just trying to learn more. And who knows. When it's all said and done, I may sound like Lynchjm II.

[lynchjm24]

Originally Posted by lynchjm24

To be honest you are wasting your time trying to get historical leagues to meet your expectations. The game can't properly import defensive ratings based of what's in the database.

This is what I said. I did not say to shut down conversations.

I'm saying quite literally that if your goal is for historical OOTP leagues to replicate closer to history some of the results pitchers achieved, it's an impossibility based on the way the game models defense. Because defense is such a huge part of the results engine and you can't get accurate ratings for defensive players based on the information in the databases, from there it's impossible to recreate their historical performances on balls in play.

If your intention is to motivate Markus to revamp the game, you are going to need to bring a large amount of convincing data. Talking about a handful of players that pique your interest isn't advancing anything.

[Le Grande Orange]

Originally Posted by RchW

A home run is strictly pitcher controlled.

Or perhaps more accurately a momentary lack of control.

[Left-handed Badger]

Originally Posted by RchW

A home run is strictly pitcher controlled. What led up to it was as described.

Originally Posted by lynchjm24

Ah, I didn't get your point either. Richmond gave up a walk-off HR to Harris in the 12th to end it.

I was into it because I like when teams run out of pitchers like Toronto pretty much had and they used this coming Tuesday's starter in relief. Although Cito said they had Hayhurst available, he didn't use him.

Right, Richmond HR was mentioned in the post. I was basically interpreting RchW's post as saying it was somehow Cecil in control of Wells not getting those balls.

ANd Lynch, I am high on defense, always have been. But, I want to know if pitchers have ANY value in sabremetrics, barring the guy who can K 1+ an inning. Or should we just pay minimum wage for 12 pitchers and invest all our money in the position players.

[Daern]

I may have missed this somewhere, but have you tried looking at BABIP variations park-by-park and year-by-year, as opposed to pitcher-in-park? It might show something about why pitchers are better at home--some parks may just be bad/good BABIP parks.

[lynchjm24]

Originally Posted by Left-handed Badger

Right, Richmond HR was mentioned in the post. I was basically interpreting RchW's post as saying it was somehow Cecil in control of Wells not getting those balls.

ANd Lynch, I am high on defense, always have been. But, I want to know if pitchers have ANY value in sabremetrics, barring the guy who can K 1+ an inning. Or should we just pay minimum wage for 12 pitchers and invest all our money in the position players.

I think in general teams that are deeply invested in sabremetrics are statistically looking for:

High K Rates
Low BB Rates
High GB Rates
Lack of a huge platoon split
Ability to be efficient


You really do need to be able to do 2 out of 'big' 3 (K/9, BB/9, HR/9) to be successful in the major leagues. Strike batters out, avoid walks and keep the ball in the park. The best do all three.

I think the market is rewarding pitchers that consistently do 2 or 3 of these things. We don't see the huge contracts going to pitchers anymore who had a good ERA but didn't have success in these areas.

On top of these skills you need to be able to stay healthy. The money is going to the guys who take the ball every fifth day. You get a premium if you can stay in the rotation. Derek Lowe's contract this off season I think was a good example of that.

If your conclusion is that people who believe in DIPS don't think pitchers are valuable, then I think they haven't gotten their message through to you, because I don't think that is the point in anyway.

[knockahoma]

I'm saying quite literally that if your goal is for historical OOTP leagues to replicate closer to history some of the results pitchers achieved, it's an impossibility based on the way the game models defense.

Consider this, Lynch. Historicals are an excellent way of fine-tuning the fictional engine. [B]Everybody wants a proper variance in BABIP in their fictionals, right?

Historicals are like artillery shells and the fictionals would be the target. Each shot "walks us in" to the target.

When I ran the Messersmith test, I was looking for too-wide variances in BABIP.

I wasn't concerned if San Diego, or Detroit were defensively imported correctly. It didn't matter. I just wanted a bad defensive team. I'd already run 3 1974 seasons. San Diego and Detroit ended up 5th, or 6th, each time. Bad defenses. I wanted to see what would happen to a good pitcher's BABIP from good team to bad team. Didn't have to be Messersmith. Could have been a fictional guy named Jim Lynch. In fact, let's pretend it was:

When Jim Lynch, a top pitcher, hurled for a good defensive team, he almost always finished with a BABIP around .230-.250. His ERA was usually 2.65 to 3.10. But when the same Jim Lynch played on a bad defensive team, even though his HRs, BB, Ks remained pretty constant, his BABIP destroyed his season. 5 out of 6 seasons, his BABIP was around .300, or above. Because of that, his ERA sky-rocketed, usually around 3.85 into the mid 4s.

So, the question I asked myself is this: Do I believe that a top pitcher's BABIP will EXPLODE going from a top defense to a poor one? Now, BABIP is very important here, because HR, BB, K number remained extremely constant in the Messersmith tests.

It was BABIP alone that took a top pitcher and threw him into the bottom 20% of starters.

Someone brought up Mike Hampton. His BABIP did rise in Colorado, but what destroyed his season was 31 homers allowed versus 10 the previous season. I do see that kind of thing quite often in history.

So far, in my looking, I DON'T see that as a real life occurence.-- certainly not a regular one, which it appears to be in X.

So, I'm looking for those examples. And this is the kind of search where you can't go through league stats. You have to take it one pitcher at a time.

If BABIP is accurately divided between influences, then pitchers may become more accurate than now-- even if the defensive import problem remains.

My struggle, and one I've been working on the past few days, is to capture individual BABIP ranges from real life pitchers and compare them to OOTP results. It's slow going.

I sense there's something there, though. Because whenever you see a "good" historical pitcher blow up in OOTP with a 4.50-5.50 ERA, look at his K's, BBs, HRs. They're probably pretty close to historical. It's the BABIP that WAY out of the park. And it's that way consistently in sim after sim!

Q: Maybe that's just luck of the draw.

A: I think it's more than that. It appears that Messersmith on a poor defensive team will turn out with a high ERA (same Ks, BBs, HRs) and high BABIP 5 out of 6 times. The influence of defense probably should be less intrusive than that.

If I had to put it in numbers (based on what I see historically, so far), Messersmith, going from a good D to a bad one, more often would see his ERA go from 2.65 up to 3.10-3.50 (I'm talking 70s numbers, here)

But, all of the questions and challenges I anticipate from you and others are going to be answered by my looking at individual pitchers, real life VS OOTP.


Summing up: Historically good pitchers blow up not because of a huge change in HRs, BB, K's, but because of BABIP. The same effect probably exists in fictionals. The effect seems far more intense in OOTP than in real life under the same conditions. If so, then Marcus, having access to real life BABIP variances, might tweak a bit here, or there, to get a better result. In fact, it might be a simple fix.


That's what I'm trying to check.

[lynchjm24]

Originally Posted by knockahoma

Consider this, Lynch. Historicals are an excellent way of fine-tuning the fictional engine. Everybody wants a proper variance in BABIP in their fictionals, right? We don't want a pitcher's luck to vary from a .188 BABIP to a .422 BABIP, for example (given a full season, that would be unrealistic, even to a fictional player).

Historicals are like artillery shells and the fictionals would be the target. Each shot "walks us in" to the target.

When I ran the Messersmith test, I was looking for too-wide variances in BABIP.

I wasn't concerned if San Diego, or Detroit were defensively imported correctly. I'd already run 3 1974 seasons and they ended up 5th, or 6th. My test was in search of a bad defense, not a properly imported one. I wanted to see what would happen to a good pitcher's BABIP from good team to bad team. Didn't have to be Messersmith. Could have been a fictional guy named Jim Lynch. In fact, let's pretend it was:

When Jim Lynch, a top pitcher, hurled for a good defensive team, he almost always finished with a BABIP around .230-.250. His ERA was usually 2.65 to 3.10. But when the same Jim Lynch played on a bad defensive team, even though his HRs, BB, Ks remained pretty constant, his BABIP destroyed his season. 5 out of 6 seasons, his BABIP was around .300, or above. Because of that, his ERA sky-rocketed, usually around 3.85 into the mid 4s.

So, the question I asked myself is this: Do I believe that a top pitcher's BABIP will EXPLODE going from a top defense to a poor one? Now, BABIP is very important here, because HR, BB, K number remained extremely constant in the Messersmith tests.

It was BABIP alone that took a top pitcher and threw him into the bottom 20% of starters.

So far, in my looking, I DON'T see that as a real life occurence.-- certainly not a regular one, which it appears to be in X.

So, I'm looking for those examples. And this is the kind of search where you can't go through league stats. You have to take it one pitcher at a time.

If BABIP is too closely tied to team defense, then even if import problems remain, the pitchers become more consistent to real life.

My struggle, and one I've been working on the past few days, is to capture individual BABIP ranges from real life and apply them to OOTP.

I sense there's something there, though. Because whenever you see a "good" historical pitcher blow up with a 4.50-5.50 ERA, look at his K's, BBs, HRs. They're probably pretty close to historical. It's the BABIP that WAY out of the park.

Q: Maybe that's just luck of the draw.

A: I think it's more than that. It appears that Messersmith on a poor defensive team will turn out with a high ERA (same Ks, BBs, HRs) and high BABIP 5 out of 6 times. The influence of defense probably should be less intrusive than that.

If I had to put it in numbers (based on what I see historically, so far), Messersmith, going from a good D to a bad one, more often would see his ERA go from 2.65 up to 3.10-3.50.

But, all of the questions and challenges I anticipate from you and others are going to be answered by my looking at individual pitchers, real life VS OOTP.

All well and good, you just have no way to know which teams in your sample have good defense and which teams don't. That's a real life problem as well as an OOTP problem.

You might be able to find some examples of pitchers who moved from good defenses to bad defenses... based on someone's opinion. If you cull every trade from the 70's how are you supposed to know who moved from a good defense to a bad defense? How do you take park effects out of the numbers? Maybe Messersmith went from a good defense to a bad defense... making that determination based on the stats you pulled from baseball-reference cannot be done. Those numbers are pretty much useless to base the overall quality of team defense on.

I agree that sure, looking at the range of BABIP over the last 30 years and having fictional leagues fall into the pitcher's range makes sense. I would guess that based on just playing OOTP that it's going to be awfully close to the real life ranges. If it wasn't
A: Someone would already be jumping up and down about it
B: The results in fictional leagues would feel off, and they don't really

Again it seems like your goal is for historical leagues to have BABIP results similar to real life. I do not think that can happen without a total rewrite of the results engine, which again you'd need volumes of compelling evidence, of which you have none.

[lynchjm24]

I honestly don't see how you struggle for examples. One of the bigger baseball stories in 2009 is that the Rangers defense is helping them win games.

Team BABIP
2008 - .317
2009 - .288

Kevin Millwood BABIP
2008 - .358
2009 - .248

2008
6.7 K, 2.6 BB, 1.0 HR
2009
5.0 k, 2.7 BB, 1.2 HR

2008 ERA
5.07
2009 ERA
2.62

I guess there can be variation in real life, without changing teams. Millwood a GB pitcher got a better SS/3B/1B and his BABIP dropped like a rock. Should I be suprised by that, or did he figure it out in his mid 30s?

[knockahoma]

All well and good, you just have no way to know which teams in your sample have good defense and which teams don't. That's a real life problem as well as an OOTP problem.

It's inference, you're right. I look at the end results. Let's say you run the Messersmith test 30 times.

If Messersmith has 25 good seasons in 30 as a Dodger, but 25 bad seasons out of 30 as a Tiger, what are our options? His Ks, BBs, HRs, remain pretty constant. Only his BABIP varies widely from LA to Detroit.

The inferences?

*BABIP in OOTP truly is tied strongly to defense. That explains why Messersmith is a stud BABIP in LA, and sucks in Detroit.

* BABIP in OOTP in NOT strongly tied to defense. The Virtual Messersmith just gets slammed with bad luck in Detroit, sim after sim after sim.

[lynchjm24]

Would you consider Carlos Silva to have blown up in 2008?

2007 ERA 4.19 33 starts
2008 ERA 6.46 28 starts

Home runs...
0.9 to 1.2
Walks...
1.6 to 1.9
Strikeouts...
4.0 to 4.1

His BABIP?

.302 to .344

[lynchjm24]

Originally Posted by knockahoma

It's inference, you're right. I look at the end results. Let's say you run the Messersmith test 30 times.

If Messersmith has 25 good seasons in 30 as a Dodger, but 25 bad seasons out of 30 as a Tiger, what are our options? His Ks, BBs, HRs, remain pretty constant. Only his BABIP varies widely from LA to Detroit.

The inferences?

*BABIP in OOTP truly is tied strongly to defense. That explains why Messersmith is a stud BABIP in LA, and sucks in Detroit.

* BABIP in OOTP in NOT strongly tied to defense. The Virtual Messersmith just gets slammed with bad luck in Detroit, sim after sim after sim.

We already know the answer to that. BABIP in OOTP is strongly correlated to defense, just like it is in real life.

[knockahoma]

Hey,

Millwood's interesting. I've got to wrap it up tonite. But, I'd like to look more at that situation tomorrow.

We already know the answer to that. BABIP in OOTP is strongly correlated to defense, just like it is in real life.

Then, that's the answer to your previous concern with regards to my Messersmith study:

you just have no way to know which teams in your sample have good defense and which teams don't. That's a real life problem as well as an OOTP problem.

Studying the same teams through 30 sims gives you a good idea of which teams are importing good and bad defenses.

But, hey. Millwood's a great example and I don't want to gloss over him. Maybe he'll bust my concerns. But, I've gotta check a few things, too.