Question about BABIP

[PhilWings24]

I know this is a pretty controversial stat, and i'm sure someone has asked this question before me.

There's something strange about the .290-BABIP-assuming-average-defense to me. I'm not challengingi it or anything, there's just one part of it that doesn't make sense.

If that stat is correct, then how come some batters that stike out alot hit 300 (manny ramirez) while some players that hardly ever strike out don't hit for impressive averages, necesarily (tougher to think of an example for this one, but i guess angel berroa works...orlando cabrera too i guess).

it seems to me that for that stat to be fully accurate, it would have to mean that any hitter hitting above .290 had gotten lucky. (i guess it would be a bit higher when you take homeruns into account, but you get what i mean)

the obvious answer to me would be that better hitters hit the ball harder, and harder hit balls fall for hits more often. but it would also seem to me that worse pitchers get hit harder, and that doesn't show up in their BABIP, only in their opponents slugging%.

so the only explanation to me would be that some hitters really do have a knack for placing the ball, or "hitting em where they ain't."

well...that's my question...basically just that if its true the only reason a pitchers BABIP should be above 290 is bad defense/bad luck, why isn't the converse true for hitters?

thanks for any help you can give me

[ssbguyincognito]

I don't think BABIP has anything to do with hitters.

The theory, if I'm not mistaken, says that a pitcher has very little control over the outcome of a ball once it is put into play -- i.e., say a batter makes contact with a pitch. At that point, whatever happens is a function of the batter and of luck.

Note that this doesn't say anything about the hitter. When Ortiz makes contact with a pitch, he has contorl over what happens to it. Where the pitch lands will be a function of his skill (such as his batting stance, his eye, his timing, his range, where he hits it on the bat, when he first swings, etc...) and of luck.

BABIP doesn't imply that whenever someone makes contact with a pitch, whatever happens from that point on is luck. Instead, it's just saying that pitchers differ mostly in three levels: walks, Ks, and HRs.

[PhilWings24]

if it has to do with pitchers, it has to do with hitters.

i get the point you're making, but intuitively it makes no sense. in fact it was my original explanation, but i'm not entirely satisfied with it: if a pitcher has no control over the outcome of a ball once it is hit into play, how does the hitter?

if david ortiz's skill as a hitter makes balls he puts into play more likely to fall, then why doesn't barry zito's skill make them less likely to fall?

if a hitter has an observable skill to hit the ball in a way that makes it more likely to land, it doesn't make sense- at least not intuitively- that a good pitcher wouldn't be able to combat that skill.

and its my understanding that the babip theory DOES say whatever happens to a ball once its hit into play is luck, or defense. i understand that there is some, very very little, evidence that a pitcher with a good BABIP will have a good BABIP next year, but it varies much more widely than a hitter's BABIP seems to. and i don't get why.

basically, what i'm saying is if a pitcher's BABIP is a product of defense and luck, why isn't the converse true for hitters?

[ssbguyincognito]

Because what's true for pitching is not necessarily true for hitting. You first statement "if its true for pitching its true for hitting" is wrong. Why should something that's true for pitching necessarily be true for hitting?

[PhilWings24]

Because what's true for pitching is not necessarily true for hitting. You first statement "if its true for pitching its true for hitting" is wrong. Why should something that's true for pitching necessarily be true for hitting?

i never made that first statment lol.

i said if it has to do with pitching, it has to do with pitching. very very different than saying if its true for one its true for the other.

a pitcher's job is to get the batter out. its the batter's job to NOT get out. therefore, anything a pitcher does has to do with what the batter is doing. i brought that point up because you said something like "because babip has nothing to do with hitters," in your first post.

and i get that something thats true for a pitcher isn't necesarily true for a hitter, but i'm asking why it isn't.

if something a batter has an observable skill of getting balls he its into play to land for hits, why shouldn't a pitcher be able to do something to try and combat that skill? i understand that there's no reason to think that a pitcher MUST be able to combat that skill, but i'm looking for an explanation as to why he can't.

i mean, try and think of one other offensive statistic that batters have exhibited consistency with, that pitchers have not done the same. hits, walks, doubles, hr, obp, slg, avg, hbp, cetain hitters and certain pitchers have shown a particular ability to either prevent or achieve each of those stats.

and i'm just looking for an actual explanation as to why BABIP is the first where that isn't true. i don't doubt that it is, in fact i believe that it is. i don't think that after having thousands of people try to disprove this, i finally found a hole in the theory after thinking about it for atotal of about 20 minutes lol. but i need an explanation as to why hitters have infinitely more say as to what happens with a ball after its hit than pitchers do.

[ssbguyincognito]

You're right, I did misinterpret your statement.

As for the reason why, there are probably a billion different reasons, each a factor that affects the game of baseball. My explanation (which I don't claim is true) is that once a batter makes contact with the pitch, anything he does from tha tpoint on is going to affect the ball's trajectory from that point forward. That makes sense.

But once a batter makes contact with the ball, anything the pitcher has done to the ball was done a long time ago (relatively speaking) and from a far distance. So the pitcher's impact on the ball by the time it hits the bat has probalby changed immensely and been affected by the almost infinite number of random variables at play (linear momentum, angular momentum, air pressure, density, humidity, temperature, spin, gradients of all of these, etc...).

I would argue that what a pitcher does to a pitch determines its trajectory going to the batter. But what the batter does upon hitting the pitch determines its trajectroy back into the field of play.

[PhilWings24]

that makes some sense....this would really be a better question for a coach of something, someone that could break down the fundmentals of a batter vs. a hitter.

just repeating myself, it seems like there is something that good hitters do that makes a ball they hit into play more likely to land for a hit, there should be something that pitchers can do from consistently limiting a hitter's ability to do that...

thanks for your help though, any ideas are useful, this is gonna bother me til i stumble upon the right answer lol...

[stevebogus]

My theory:

There is a much larger variation between different batters than there there is between different pitchers.

Why?

Pitchers aren't selected on anything but their skill as pitchers. Unskilled pitchers don't make it to the majors, so the quality level of pitchers is compressed. But batters (except for a DH) are two-way players, and some fielding positions are so difficult that teams cannot find good hitters to play them. If baseball went the way of football and allowed unlimited substitutuion you would see lineups filled with good hitters, with less variation between them.

It isn't than major league pitchers have no skill at controlling BABIP. It is just that they all have approximately the SAME skill. As soon as they lose a little bit of that skill they are no longer major league pitchers.

[PhilWings24]

My theory:

There is a much larger variation between different batters than there there is between different pitchers.

Why?

Pitchers aren't selected on anything but their skill as pitchers. Unskilled pitchers don't make it to the majors, so the quality level of pitchers is compressed. But batters (except for a DH) are two-way players, and some fielding positions are so difficult that teams cannot find good hitters to play them. If baseball went the way of football and allowed unlimited substitutuion you would see lineups filled with good hitters, with less variation between them.

It isn't than major league pitchers have no skill at controlling BABIP. It is just that they all have approximately the SAME skill. As soon as they lose a little bit of that skill they are no longer major league pitchers.

that's an interesting idea...but then that would still beg the question, how come they seem to have so much less control over babip than any other statistic?

the consistency of a pitcher's ability to control his babip isn't nearly as apparent as a pitcher's ability to consistently control HR allowed, 2b allowed, avg against, obp against, etc.

[538280]

that's an interesting idea...but then that would still beg the question, how come they seem to have so much less control over babip than any other statistic?


Simple answer-the defense has a ton (probably more than the pitcher himself) to do with whether or not balls in play are hits or outs. The other things like BB/9, K/9, HR/9, and such are basically purely the responsibility of the pitcher, so it's natural they'd be more consistent.

[stevebogus]

that's an interesting idea...but then that would still beg the question, how come they seem to have so much less control over babip than any other statistic?

the consistency of a pitcher's ability to control his babip isn't nearly as apparent as a pitcher's ability to consistently control HR allowed, 2b allowed, avg against, obp against, etc.


Pitchers have less control over BABIP because they depend on fielders to make those plays. Strikeouts and walks are mostly a pure pitcher vs. hitter battle. Homeruns too. But once a ball is put in play other players get involved, and the result is, to a large degree, determined by the skills of the fielders.

[ssbguyincognito]

Actually I remember that HRs are very erratic from season to season as well (the correlation for HRs given up by a pitcher was around 0.27). This makes sense: pitchers change leagues, switch teams, go to different ballparks, get injured etc...

I also remember reading another article called DIPS 2.0 by Voros. He said that, after more rigorous studies were down, it was shown that pitchers do have a very limited control on BABIP -- there are some pitchers with consistently low BABIPs. For example, knuckleballers have very low BABIPs, and that's probably because the knuckleball does change how well it can be placed for a hit. But he also maintained that pitchers vary very slightly in terms of BABIP, and very largely in terms of walks, Ks, and HRs.

[ssbguyincognito]

Pitchers have less control over BABIP because they depend on fielders to make those plays. Strikeouts and walks are mostly a pure pitcher vs. hitter battle. Homeruns too. But once a ball is put in play other players get involved, and the result is, to a large degree, determined by the skills of the fielders.
That's a good point. If that were true, then we would expect different pitchers on the same team might have BABIPs close to each other (because they share the same defense). Has anyone heard of or done any research about this?

[Mariano_Rivera]

Actually I remember that HRs are very erratic from season to season as well (the correlation for HRs given up by a pitcher was around 0.27). This makes sense: pitchers change leagues, switch teams, go to different ballparks, get injured etc...
Not to mention moving to the Yankees and benifiting from Melky Cabrera :laugh

[Sliding Billy]

My theory:

There is a much larger variation between different batters than there there is between different pitchers.

Why?
.....
But batters (except for a DH) are two-way players, and some fielding positions are so difficult that teams cannot find good hitters to play them. If baseball went the way of football and allowed unlimited substitutuion you would see lineups filled with good hitters, with less variation between them.

Interesting theory. BABIP tracks BA pretty closely (duh :rolleyes: ). However, there is also variation on BABIP within groups of good hitters and within groups of bad hitters. Whether it's greater than the variation within groups of good pitchers and within groups of bad pitchers, I do not know.

As an extreme example take Joe Dimaggio and Rod Carew:

-------AB--HR--SO--BIP---H---HIP--BA--BABIP
Joe D 6821 361 369 6091 2214 1854 .325 .304
Rod C 9223 92 1028 8195 3053 2961 .328 .361

"A relatively high BABIP" = "A relatively low HR/K ratio"

[Tango Tiger]

I suggest reading the best research on the subject here:

http://www.tangotiger.net/solvingDIPS.pdf

And this research is also useful:
http://www.tangotiger.net/dipsbands.html

SteveBogus is the one on the right track.

[rluzinski]

And this research is also useful:
http://www.tangotiger.net/dipsbands.html

So, pitchers who are allowed to accrue a large amount of IP are more skilled, including with regard to their ability to suppress hits on balls in play. Is that a fair conclusion or is there another selection bias I'm over looking?

On a side note, why doesn't SO rate correlate with BABIP at all? Should it? Do the other factors drown out it's contribution?

[Tango Tiger]

You'd be hard-pressed to find any two things that are in the same realm to have a correlation coefficient of zero. Can you reword your question, using the actual r?

As for the article, like I said, I didn't want to make any conclusions.

[PhilWings24]

man my computer is flipping out today...accidentally double posted, sorry...

[PhilWings24]

[I]
On a side note, why doesn't SO rate correlate with BABIP at all? Should it? Do the other factors drown out it's contribution?

hahah that side note, is more or less what i'm asking...the only explanation to it is that (other than other factors drowning it out, i suppose) is that hitters somehow have more control of what a ball does after contact is made than a pitcher does, which makes no sense.

it makes, some sense that the hitter's speed would be the reason, but even that makes less and less sense the more i think about it. if speed is all there is to it, angel berroa should be doing great, papi's average should be somehwere around .013 lol, and better pitchers should be able to control that by jamming the hitter; making him more likely to pop it up.


Pitchers have less control over BABIP because they depend on fielders to make those plays. Strikeouts and walks are mostly a pure pitcher vs. hitter battle. Homeruns too. But once a ball is put in play other players get involved, and the result is, to a large degree, determined by the skills of the fielders.

that makes perfect sense to me, but what i'm asking is why isn't that the case for a hitter?

the reason i asked the question you were answering is that: if some hitters have a particular knack for getting doubles, some have a particular knack for hitting homeruns, and having high BABIPs, why is it that pitchers have a particular knack for preventing doubles, or a particular knack for preventing homeruns, but practically no control over their opponent's BABIP?

what is it about BABIP that makes it the only thing about a batter that a pitcher can't defend against?

(and sorry if one of those links perfectly addresses that question lol, i just got online for the first time today and was eager to write that post, and i'm yet to read what's in the links. thanks again for your time, everybody)

[Tango Tiger]

Read the links.

[SABR Matt]

Batters have the ability to control their BABIP far more than pitchers do...some batters (line drive hitters and very fast baserunners) can strike out a lot and hit .300.

But even ignoring the positive benefits of hitting ball in play line drives, you're forgetting that HRs count on your BA. As long as you don't strike out way WAY more than you homer, you can hit .300. Think of it this way...if 30% of you non-ball-in-play at-bat producing events are HRs you should hit .300 in an average year.

[PhilWings24]

Batters have the ability to control their BABIP far more than pitchers do...some batters (line drive hitters and very fast baserunners) can strike out a lot and hit .300.

But even ignoring the positive benefits of hitting ball in play line drives, you're forgetting that HRs count on your BA. As long as you don't strike out way WAY more than you homer, you can hit .300. Think of it this way...if 30% of you non-ball-in-play at-bat producing events are HRs you should hit .300 in an average year.

the HR theory did occur to me recently, but (though i haven't tested it), that doesn't seem like it should explain it, at least not intuitively.

Jeter, in balls he's hit into play, is
157/408= .3848

Last Year
183/518= .3532

2 Years Ago
165/521= .3167

And yes that is a useless sample size, but just by glancing at his numbers, i can tell the majority of his BABIPS are in that general area. And i'm assuming that jeter isn't a freak of nature, and there are other players like him. not exactly the scientific way to go about it, but it works for the sake of discussion.

So i don't think the HR's account for it fully. Speed makes some sense, but again, not entirely. Manny Ramirez comes to mind there; he's had some great batting averages in years where he's only hit 30 some homeruns. Also, Jeter-while certainly fast- is no joey gathright and no carl crawford.

And i have read the links, and i don't think they adress the question. The line drive theory makes sense to me, but that seems to disprove the BABIP theory.

There's no question bad pitchers get hit harder than worse pitchers (at least i've never understood there to be one). However, the BABIP theory states that the strength with which a ball is hit has little effect on the percentage of the time it lands. it shows up in the slugging percentage, and so on and so forth, but not the batting average. At least that's my understanding of it.

So then, if good hitters have consistently high BABIPs because they hit line drives more often than a bad hitter, why don't good pitchers have consistently low BABIPS because they prevent line drives more often than a bad pitcher?

[SABR Matt]

The answer to your final question is that good pitchers have lower BABIP than .300 but it is not as far from average (there is less range form good to bad pitchers) because the pitcher cannot react to the batter in the middle of a pitch whereas the batter can read the pitch as it's coming in (at this point the pitcher can't do anything about the ball...it's out of his hand), adjust his approach to the ball and control its' path with a little more regularity. The batter has more information when he takes his action to contribute to a ball in play event than a pitcher does when he takes his.

[Mariano_Rivera]

Post Deleted

[PhilWings24]

The answer to your final question is that good pitchers have lower BABIP than .300 but it is not as far from average (there is less range form good to bad pitchers) because the pitcher cannot react to the batter in the middle of a pitch whereas the batter can read the pitch as it's coming in (at this point the pitcher can't do anything about the ball...it's out of his hand), adjust his approach to the ball and control its' path with a little more regularity. The batter has more information when he takes his action to contribute to a ball in play event than a pitcher does when he takes his.

that makes, by far, the most sense of any explanation i've heard.

it alone doesn't 100% answer my question, but it just about does, and i think i've figured out at least enough to keep this from nagging at me for a few days...

basically, as i've tried to articulate but haven't done a very good job, what confused me about all this is the fact that for every offensive skill i could think of that has been proven to be an actual, observable, repeatable (so to speak) skill has a counterpart that can be exhibited, except this babip stuff.

but it makes sense (and the more i think about it this is more or less exactly what you said, SABR Matt, i'm mostly typing it out myself to put closure on the subject lol) that the skill that makes a a batter have a consistently high BABIP is too fine and too subtle for a pitcher to really defend against. that a what gives a hitter a good babip is the ability to slap tough pitches the other way, all that good scoutastic stuff, that a pitcher can't really influence with by making good pitches as much as he can with things like homeruns and walks allowed.

hm...i wonder then if there would be reason to value a batter's babip, at least a little...and i know this would be a microscopic difference, but if you had 2 players of overall value equal value, but one had a higher BABIP, i wonder if he'd be more prone to succeed in the playoffs, against presumably better pitching. like i said though, i'm sure if that matters at all, it barely does. and i sort of doubt that it does matter at all

thanks again for your time and walking me through all this haha

[SABR Matt]

Watch Ichiro hit and you'll understand why some batters have the BABIP skill.

Yeah some of his high BABIP is caused by all of his infield hits. :D But when he's going well, you'll see him react to a pitch mid-flight, change the position of his hands or take a little crow hop if he needs to reach out for a pitch and line it up with his eye like it's coming to him in slow motion...

He says what he does is imagine the pitch is strung on a clothes line...the line pierces the ball at the center of its' rotation and the ball must follow that imaginary path...he makes it his mission to hit the ball right back through the same path in the other direction...the result is a lot of low screaming line drives to left center and center and a BABIP higher than .300 even once infield singles are removed.

[rluzinski]

You'd be hard-pressed to find any two things that are in the same realm to have a correlation coefficient of zero.

I should have said, "statistically significant correlation". I based that off of data from "The Effect of Batted Ball Types on Balls in Play (http://bradbury.sewanee.edu/Balltypes.pdf#search=%22babip%20STUDY%22)" by J.C. Bradbury. Now, he only used a half a season's worth of data, so I wouldn't be suprised if there was a more comprehensive study out there.

[Sliding Billy]

Watch Ichiro hit and you'll understand why some batters have the BABIP skill.

That's an excellent example, and while I don't want to muddy the clearing waters once again, it seems to me that BABIP is kind of a residue stat, what's left after you take away K's and HR's, and that helps make it elusive.

Players who rarely strike out have good bat control, but if you take two players with comparable BA's and power, the one who strikes out more will have the higher BABIP. Power is a constituent of hitting ability, but if you take two players with comparable BA's and K rates, the one who hits fewer home runs will have the higher BABIP.

To me, BABIP isolated from general batting ability seems more a characteristic of certain hitters, rather than a specific proficiency.

[SABR Matt]

I would certainly agree there. You don't look at BABIP as a specific hitting skill "when he puts it in play and doesn't homer he's really good at doing X"...that would be silly. It's a predictive tool when you're looking at either hitters or pitchers, becasue when a player's BABIP is significantly different than you normally expect given a player's typical batting style, it's a probably a sign that he was lucky or unlucky and you can adjust his future projections accordingly.

An example I remember from way back in 2002. Dan Wilson (Mariner catcher) was catching all kinds of praise for his unusually high batting average that year. Announcers and Ms fans all over the net were wondering if we were premature in dismissing Wilson as washed up. I was not wondering...I was CERTAIN he was still washed up, and I proved it by looking at componant offensive statistics including BABIP:

Year BABIP ISO BB/PA K/BB
1996 .317 .159 .059 2.75
1997 .290 .153 .069 1.85
1998 .281 .142 .065 2.33
1999 .318 .116 .063 2.86
2000 .274 .101 .073 2.32
2001 .302 .138 .049 3.45 (red flag!)
2002 .368!! .101 .046 4.50 (danger Will Robinson!!)

You've got a player with skyrocketing K/BB, a sudden drop in walk rate (indicating a loss of bat speed), an almost unbroken string of decreasing power rates and no baserunning speed to speak of and somehow he's batting .295 despite HORRIBLE K rates...the BABIP is off the charts and it's pure unadulterated luck. The next season he hit .241 :)

[rluzinski]

Tango,

I understand that the pitcher has only so much control of BABIP. According to "Solving DIPS", it's about 28% with 700 BIP. If I only look at pitchers with 1000 or more IP, that gives me a sample of at least 3000 BIP per pitcher. At 3000+ BIP, the "luck" contribution is roughly halved (STD~.008), giving the pitcher about a 40% influence on BABIP (unless I've slaughtered your equations, of course).

That seems to be a big enough pie of BABIP as to be able to get a meaningful correlation if I can find what metrics capture the skill pitchers have to supress hits on balls in play. Common sense suggests that a pitcher who can make batters miss the ball altogether will also cause batters to not make make solid contact as often when they do get the bat to the ball. I pretty much already know I'm wrong but away I go...

Using the Lahmen database, I looked at pitchers who started their careers no earlier than 1975 and had acumulated at least 1000 IP through 2005. I estimated BABIP with:

BABIP = (H-HR)/(2.9*IP -.966*SO+H-HR)

The resulting R^2 between SO/9 and BABIP is .0032. Using SO/BF is even worse. The graph is a lifeless blob:



I don't expect a huge correlation but that's basically zero. Either I'm wrong, the effect is too small or there's other factors countering the SO effect. One that comes to mind:

More SO = harder thrower = harder hit balls when contact is made = higher BABIP.

Just me thinking out load, really.

[misterdirt]

Common sense suggests that a pitcher who can make batters miss the ball altogether will also cause batters to not make make solid contact as often when they do get the bat to the ball.

Although this seems like it should be true it is not. Strikeouts are a function of the number of pitches that a pitcher does right. For those pitches your reasoning might be true. However, most hits are on pitches that are mistakes. The number of mistakes a pitcher makes is not necessarily correlated with the number of good pitches he makes. When a pitcher makes a mistake whether it becomes a hit or not depends somewhat on the skill of the batter when the mistake is made but is mostly chance. Hence, the low correlation between BABIP and a pitcher's K's.

[SABR Matt]

You're correlating the wrong thing.

If you want to see whether pitchers who strike batters out a lot also have a negative impact on ball in play results, you need to correlate K rate vs. TEAM-RELATIVE BABIP...you're not getting a correlation there because the pitchers' BABIP is mostly the team defense they're in front of...their actual impact on BIP results is much smaller than the fielders behind them and so what you're actually trying to correlate in that graphic is team defense vs. K rates which is just plain ole' silly.

[misterdirt]

you're not getting a correlation there because the pitchers' BABIP is mostly the team defense they're in front of...their actual impact on BIP results is much smaller than the fielders behind them

That is not the conclusion that Tango et al. came to in the Solving DIPS article. They concluded that pitching not fielding had the bigger impact on BABIP but that both were much less than luck.

Matt - How are you calculating Team Relative BABIP?

[SABR Matt]

I haven't read the entirety of the solving DIPS article, but I suspect they're talking about correlations which I don't think is the best way to answer that question even with a large dataset.

And team-relative BABIP can be calculated one of two ways:

Pitchers BABIP / Team BABIP -- or -- Pitcher's BABIP / ((Team - Pitcher BIP Hits) / (Team - Pitcher BIP))

[misterdirt]

I haven't read the entirety of the solving DIPS article, but I suspect they're talking about correlations which I don't think is the best way to answer that question even with a large dataset.

No, they weren't.

[Tango Tiger]

I'll reiterate: read the links.

rluz: the correlation we want to look at it K/BFP and BABIP. You reported something akin to zero, even with a large number of BFP per pitcher. Interesting.

Even though you controlled for startYear being at least 1975, I still would like to see it league-adjusted. The BABIP has gone up since around 1994, and of course, so has HR and K. At the least, you should adjust for the league-year (or, maybe look at 1973-1992 instead, where the run environment is more stable, for each league).

No need to limit based on "startYear". Just look at the seasons.

[SABR Matt]

I still think it's wrong-headed to try to correlate K rate and BABIP without accounting for team defense.

[Tango Tiger]

I don't see it as wrong-headed, since you are simply ignoring a variable. Your regression will simply be lower.

On the other hand, given enough years, say a pitcher's career, given enough of a team change over, or switching teams, the "true fielding talent" of the players behind a pitcher, will automatically converge towards zero. (How close to zero depends on everything I just said.)

[SABR Matt]

Possibly true...most of the time.

Tom Glavine's true fielding talent behind him was not zero...he got about a 6% boost.

[Tango Tiger]

I did say:
- number of years
- change in personnel
- change in teams

I should also add
- change in parks

In Glavine's case, he was stuck in the same team almost his whole career. Jones and Jones have been there the whole time. Likely Atlanta has had less turnover than a typical team.

Glavine is likely the exception that proves the rule.

***

The observed standard deviation in runs allowed in baseball is 0.59 runs per game, over the last 40 years or so.

var(obs) = var(pit) + var(field) + var(park) + var(random)

var(random) I'll guess is around .22^2 runs. var(park) I'll guess is around .15^2.

That leaves var(pit) and var(field) to total .53^2.

If var(pit) is .38^2, var(field) is .37^2 (i.e., 50%).
If var(pit) is .45^2, var(field) is .28^2 (i.e., 62%).
If var(pit) is .50^2, var(field) is .18^2 (i.e., 74%).

So, you have to figure out how much variance there is in fielding. My estmate is much closer to .18^2 than it is to .28^2.

[ohms_law]

you see, this is exactly the sort of issues that I beleave that I've been addressing. Understanding that the average major league pitcher actually allows ~ .700 BIP gives everyone a firmer basis to work with.

[rluzinski]

229 pitchers acumulated at least 1,000 IP between 1973 and 1992:



Nadda.

[SABR Matt]

Yeah...that graph is not new to me...team defense is overwhelmingly the larger factor in determining pitcher BABIP.

[rluzinski]

Yeah...that graph is not new to me...team defense is overwhelmingly the larger factor in determining pitcher BABIP.

As has already been mentioned in this thread, that opinion is contested by the conclusions found in "Solving DIPS":


So can we say that when a starter has 700 BIP, the influence on those BIP as a group can be broken down by:

luck: 44%
pitch: 28%
field: 17%
park: 11%

Solving DIPS (http://www.tangotiger.net/solvingdips.pdf)

Since I've set the IP threshold to a minimum of 1,000 innings, the "luck" piece of the pie shrinks, increasing the pitcher's influence on BABIP to about 40% (if I'm doing it right). If "Solving DIPS" is correct, that seems plenty large enough to try and find possible correlations. Of course, I set out assuming that strikeouts don't correlate to a pitcher's BABIP, so these results are not surprising to me. Just for different reasons than you.

The conclusions from "Solving DIPS" can certainly be questioned but you seem inclined to ignore them altogether. I shall do as others already have and suggest you finish reading it.

[SABR Matt]

I have read it...and I'm not IGNORING...the conclusions per say, I am bothered by the finality of the conclusion Tango et al. draw considering their primary method for reaching that conclusion is mere empiricism and not based on a logical physical model. I can certainly buy that luck plays a larger role than defense...I don't buy at all that pitching skill plays a larger role than defense...if that were the case we would see many more examples of pitchers defying DIPS-based projections year after year after year than we actually do.

[rluzinski]

I am bothered by the finality of the conclusion Tango et al. draw considering their primary method for reaching that conclusion is mere empiricism and not based on a logical physical model.

What's wrong with using empirical data to create a "logical physical model"? With that approach, a reader at least knows that the author(s) aren't simply steering their research in the direction of their choosing.

You seem to be rejecting their conclusions simply because they don't agree with your own, preconceived notions. You might prove to be 100% correct but you aren't going to convince me of it when you simply state your opinion as fact and then offer no empirical evidence in support of your claim. Show me WHY you think defense is "overwhelmingly the larger factor in determining pitcher BABIP".

Furthermore, you offer no specific explanation for why you disagree with Tango's findings in the first place. On the contrary, your prior comments suggest that you've yet to even give "Solving DIPS" a critical and thorough reading. You need to levy specific criticism of their actual work, not just dismiss it out of hand.

I don't buy at all that pitching skill plays a larger role than defense...if that were the case we would see many more examples of pitchers defying DIPS-based projections year after year after year than we actually do.

There could very well be too much statistical uncertainty with one year's worth of data. Tango's conclusions would seem to suggest as much. As I'm sure you know, the studies that look at the whole career of a pitcher has had much greater success at cutting through the fog of statistical uncertainty.

[Tango Tiger]

Since Luzinski used a 1000 IP cutoff, this would mean that he's got lots of pitchers playing on multiple teams, or even same team but different fielders. That is, the effect of park and fielders has also been reduced, given even more impact to the pitchers.

***

I looked at pitchers who pitched on multiple teams here:
http://www.insidethebook.com/ee/index.php/site/comments/and_in_this_corner/
which takes care of the fielders/park issue

See the post that starts with:

Here’s a quick study I did. I took all pitchers:
- from 1946-2005
- who played on more than one team the same season
- both in the same league
- allowed at least 50 BIP

As well as a followup post the next day.

***

And, this chart is always handy:
http://www.tangotiger.net/dipsbands.html

[SABR Matt]

It's not true that simply because a pitcher played on more than one team, the fielding issue is "taken care of"...that's a brutally oversimplified assumption. And I don't have a "preconceived notion"...I have my own studies on the influences of pitchers on ball in play results, though I wouldn't call them conclusive either since I didn't deal with things like colinearity. And I did enter this discussion with the intention of forcing anyone to believe what I do...I simply expressed my lack of confidence in "Solving DIPS" as the be all and end all question-answering final truth that it is often treated as around here.

[rluzinski]

I simply expressed my lack of confidence in "Solving DIPS" as the be all and end all question-answering final truth that it is often treated as around here.

I don't think anyone believes that it is.

[SABR Matt]

Then why is everyone here pointing to that and saying I'm "wrong" when I express the current opinion that team defense has more of an effect on pitching BABIP on an individual season basis than any other non-luck variable?

[rluzinski]

Then why is everyone here pointing to that and saying I'm "wrong" when I express the current opinion that team defense has more of an effect on pitching BABIP on an individual season basis than any other non-luck variable?

I can't stress enough that "Solving DIPS" could be completely wrong and your opinion could be completely correct. Nothing here has been "proven" right or wrong. I just don't see how you could expect anyone to agree with your position, based only on the information in this thread.

You've presented your opinion as fact, yet offered no evidence (perhaps you have somewhere else?) supporting it. Considering that evidence to the contrary has already been presented in this thread, it's difficult not to point that out every time you seem to overlook it. You also made it clear that you have not given "Solving DIPS" a critical reading, yet dismiss it's findings. That's a recipe for a very uncompelling argument.

I have no horse in this race. If you have research supporting the theory that: "team defense is overwhelmingly the larger factor in determining pitcher BABIP", I'd love to see it. I'm sure I'm not the only one.

[Tango Tiger]

It's not true that simply because a pitcher played on more than one team, the fielding issue is "taken care of"...that's a brutally oversimplified assumption.

No it isn't. If I had only a handful of pitchers, you'd be right. In the study I linked to, I have 592 pitchers who played on multiple teams in the same season and in the same league. That essentially wipes out the effect of the parks and fielders from the correlation.

[SABR Matt]

No it doesn't Tango...and this is exactly why I don't trust your study. A) 592 pitchers isn't enough to make any claims about the sample size having a significant "centering" effect on the correlation and B) correlations even of that magnitude are highly sensative to changes in even a few data points...there are countless examples of pitchers who pitched on two teams but they both happens to be pitcher's parks...or they both happened to have great defenses...or the reverse.

If even 50 of your 592 pitcher points are poisoned in ways you're ignoring, your entire study is blown up.

[Tango Tiger]

Of course you'll have pitchers in pitchers parks or hitters parks on multiple teams. The key is if such a distribution is essentially random. If say 20% of parks are pitchers parks and 20% are hitters parks, then in my group of 592 pitchers, I'd expect 4% of those pitchers to go from pitcher park to pitcher park (24 pitchers), etc, etc. Same deal with the quality of the fielders. This is why we look at pitchers who played on multiple teams.

Unless you believe that a disproportionate number of pitchers a traded from pitcher-to-pitcher park, or great-fielding-to-great-fielding teams, then we're safe to assume a random distribution in the change of the parks and fielders.

As for "highly sensitive to a few data points", that's the case for any correlation you can produce with 592 data points! When I ran with 864 data points, the standard error was .036, meaning that my correlation (r) was .18 +/- .07. At 592, the SE was .043.

What standard error would you accept?

[Tango Tiger]

Ok, I checked. The correlation in park factors was r-squared of .004. Since I reported a correlation of the pitchers' BABIP of .18, then this would go down to .17 after accounting for the non-randomness, such that it is, of the park factors.

***

If I exclude anyone who was in a hitter's or pitcher's park in either of the multiple teams, the "x" (see my earlier link) in my regression equation becomes 662, which is pretty much what I've been reporting without these exclusions.

[SABR Matt]

Of course you'll have pitchers in pitchers parks or hitters parks on multiple teams. The key is if such a distribution is essentially random. If say 20% of parks are pitchers parks and 20% are hitters parks, then in my group of 592 pitchers, I'd expect 4% of those pitchers to go from pitcher park to pitcher park (24 pitchers), etc, etc. Same deal with the quality of the fielders. This is why we look at pitchers who played on multiple teams.

Unless you believe that a disproportionate number of pitchers a traded from pitcher-to-pitcher park, or great-fielding-to-great-fielding teams, then we're safe to assume a random distribution in the change of the parks and fielders.

As for "highly sensitive to a few data points", that's the case for any correlation you can produce with 592 data points! When I ran with 864 data points, the standard error was .036, meaning that my correlation (r) was .18 +/- .07. At 592, the SE was .043.

What standard error would you accept?

The problem I have is that a lot of pitchers get "labeled" by GMs. Jarrod Washburn would never sign with the Texas Rangers or the Colorado Rockies because both Washburn and the GMs of teams with parks and defenses disfavorable to flyball pitchers know that it would be disasterous for Washburn to do so. There are a number of examples of this...teams going after only pitchers they think will fit their gameplan (Rockies and groundballers, Yankees and power pitchers, Braves and contact/groundballers etc)...the GMs aren't blind...they know there are reasons to target certain types of pitchers. It cannot therefore be assumed that pitcher X has all 30 parks to choose from and will choose from those parks randomly.

[Tango Tiger]

I pre-replied to your point here by only focusing on pitchers who pitched in neither hitters nor pitchers parks.

[SABR Matt]

OK...now deal with the team defense effect...check team DERs and see what happens to your correlation because as you recall, I didn't say parks were a bigger factor than team defenses...quite the contrary.

I *know* that pitchers experience a wide variety of team defenses because the career team BABIP when pitcher X is not pitching...even for pitchers who change teams a lot...has a fairly large standard deviation (what I mean by that is that if you take any pitcher...subtract his BIP hits and balls in play from his team totals in each season...and calculate a career team BABIP...you're going to find a large range of such values even for pitchers with long careers...I know this, because I've calculated this).

[Tango Tiger]

Your second paragraph is this:
http://www.tangotiger.net/dipsbands.html

[Tango Tiger]

Good call on the team DER, Matt!

I now limited my sample of pitchers to where the team DER (with the pitcher in question removed) between the two teams is within .010 (for an average difference of .005).

The "x" in the regression equation is now 1660 (meaning we expect r=.50, when BIP = 1660). Therefore, when BIP = 700, r=.30. That is, 700/(1660+700)=.30.

On page 18-19 of Solving DIPS, we said we r=..... .28.

We already know that luck accounts for .44.

The rest is fielding and park.

[SABR Matt]

OK this is fascinating...

Your pitching bands chart is eye opening to me in that the obvious trend is that pitchers who have long careers are more likely to defy DIPS (by defy I mean consistantly step outside the standard error range), which tells me that perhaps there is something to this notion that pitchers have more of an influence on the BIP events that I have believed because the ones who prevent hits stick around and the ones who don't...don't. It's all the weaker pitchers who don't stick that are causing the noise that obstructs some of my previous research perhaps.

So your conclusion when you limit pitchers to the ones who pitch on teams whose' DERs don't change much is that the pitchers themselves are pulling 30% responsibility for BIP results and the rest is luck...this being for pitchers who stick aroudn for a long time, pitch full seaosns with different teams etc. In other words, we can expect that the good pitchers will do more to take control of the BIP results than the bad ones.

Perhaps rather than accepting one number based on one population (one that doesn't do a very good job matching the actual population of major league pitchers) the conclusion to draw from this is that the better you are, the more we can trust you to repeat your BABIP...the worse you are, the more you're at the mercy of your teams and parks?

[baseballPAP]

You guys have my head spinning....but....

What if the reason the "bad luck" pitchers with shorter careers have shorter careers is just that....they can't stick around because those dinkers and bloops cause their numbers to be bad, and thus their careers to be short.

I do realize that that is statistically impossible to prove, but the paradox is pretty evident.

[SABR Matt]

Certainly that is possible in some cases...guy comes up with promise and potential...gets zero help from his defense and has some bad luck and runs 2 seaons of 6 ERAs and you never see him again even though it's not his fault...it happens...just ask Kenny Cloude.

I suspect bad pitchers allow more BIP hits too though...there's some cross contamination going on there.

[misterdirt]

What if the reason the "bad luck" pitchers with shorter careers have shorter careers is just that....they can't stick around because those dinkers and bloops cause their numbers to be bad, and thus their careers to be short.


That's precisely why coaches, scouts and GM's make decisions on more than just a pitcher's "numbers". And why a team with a good scout and a good pitching coach can pick up a struggling pitcher that they think has a fixable flaw that has been ignored by his past team and turn him into a productive major league pitcher.

[SABR Matt]

If you talk to anyone in baseball who's been around long eonugh Mr. D, you would hear over and over again that by in large, pitching coaches are useless and make little to no difference in the effectiveness of a pitcher. There are two or three exceptions in the game right now...Rick Peterson, Leo Mazzone, and maybe one or two more...that really make an impact. That said, teams routinely pick up "struggling" pitchers who are actually not struggling or who have a problem with pitch selection that can be fixed with a trip to the video room and some positive reinforcement...

[Ubiquitous]

So then a player out of high school is fully formed? Coaches do more then just teach mechanics and while yes some may not be all that good they are far from useless.

[SABR Matt]

I should carefully clarify here...pitching coaches are more or less useless (most of them anyway) after a pitcher has physically matured (around age 26/27). Obviously your minor league pitching coaches are critical to the development of a quality arm factory at low levels, but once you get to the big leagues, there is little that can be done to make the real effect everyone imagines.

[baseballPAP]

I have to agree with Matt here. I would say that by the time a pitcher is 27 or so, he has learned about all he is going to about pitching. The exception would be Roger Craig teaching the splitter and situations like that where a new pitch or an injury lead to a revamped style. Pitcher's who aren't forced to change somehow don't change much after their 3rd or 4th season in the bigs. There are a few good ones out there, but IMO, pitching coaches are largely the scapegoats that GMs use when their manager can't manage his pitching staff very well.

[SABR Matt]

And pitching coaches are rarely the ones teaching pitchers how to throw new pitches. That usually comes from other pitchers...Randy Johnson turned his career around when he spoke to Nolan Ryan about staying over his center of gravity (helped refine his control)...Jamie Moyer learned his circle change from another lefty whose name escapes me now...Maddux has taught his 2-seamer to several other Braves back in his glory days...things like that. Mariner prospect Mark Lowe learned a martial-arts-esque (consistent with aikido) delivery style last fall from Mariners minor league instructors and his fastball went from 93 with no command to 100 with excellent command overnight, so at a young age a coach can help propel you forward, but that is the limit of their effectiveness.

[Ubiquitous]

Coaches don't just teach they manage the resources that they are in charge of. A pitching coach has more then just one function, they have more duties then just teaching youngsters a pitch or mechanics.

[SABR Matt]

That's fine, but I wasn't talking about their job being entirely meaningless...I was talking about them being useless for changing how a pitcher approaches his craft...for making any kind of real impact on mature pitchers.

[misterdirt]

If you talk to anyone in baseball who's been around long eonugh Mr. D, you would hear over and over again that by in large, pitching coaches are useless and make little to no difference in the effectiveness of a pitcher.

I should carefully clarify here...pitching coaches are more or less useless (most of them anyway) after a pitcher has physically matured (around age 26/27).

Oh, gosh Matt thanks so much for teaching me this stuff. So now that you've back pedaled and admitted that a: there are maybe 3 or 4 good pitching coaches that can teach pitchers new stuff and b: that pitching coaches can be useful for pitchers up to age 26 or 27, i.e. the first 8 years or so of their professional career, maybe you'll go back and look at my original post and see that I made no mention of how old the struggling pitcher was or whether or not he was an established major leaguer or not. Your player that had his two seasons of 6 plus ERA's and been shipped out could have had them by age 24 or 25. You are such an idiot sometimes.

[SABR Matt]

Mr DS...I don't know what your problem is with me, but you are on my ignore list...when you grow up and start speaking in a civil manner, perhaps I'll reconsider.

[SABR Matt]

And for those of you who are still reading this ridiculous conversation aside from Mr. D, I have not backpedaled one iota. The pitching coach, with a few exceptions (and you'll note that in my original post, I said BY IN LARGE and made that exception already), generally cannot take a major league pitcher who is not effective and make him effective just by making the simple fixes that Mr D claims...when you are building a player up in the minors, it is one thing, but generally once they reach the show, the instruction is limited to confidence boosting and pre-season mechanical tuning and very few pitching coaches make much of a difference in the grand scheme of things.

I sincerely hope the rest of this community isn't fooled by Mr. D's senseless personal attacks and posturing.