Tango on Baseball Archives

ALCS Game 7 - MGL on Pedro and Little (November 5, 2003)

MGL's take on the "obvious" tiring of Pedro.

--posted by TangoTiger at 12:16 PM EDT

Posted 4:34 p.m., November 5, 2003 (#1) - RossCW
"there is evidence that pitchers do NOT have good and bad days"

Unlike everyone else in the world. And no, I don't think that is taken out of context. I think it is indicative of a world view that has become far too prevalent in sabermetrics - what I wish to be true is true.

That said, his basic point that it was not obvious Pedro was done is obviously correct. Little had seen Pedro pitch a lot and it wasn't obvious to him. I don't know how it could be to commentators in a booth or fans watching the game on television.

Posted 4:52 p.m., November 5, 2003 (#2) - dlf
Two things:

1. Without biting off all of RossCW's comment regarding the prevalency of particular attitudes among sabermetric followers ... MGL's post reads as if there is no place for subjective observation of actual performance. While I think that too much emphasis can be placed on outcome rather than process in determining good/bad decisions, I think ignoring outcome takes the exercise out of the real world and limits its applicability to the sterile laboratory where computers simulate everything.

2. Maybe it wasn't "obvious," but watching the game, I felt Pedro was done after reaching back for more to get Soriano in the 7th. Like the other poster on the fanhome thread, I thought Pedro's demeanor was that of a tightrope walker asked to cross the ledge one time too many. Also informing my view is Pedro's history of poorer performance with higher pitch counts, his frequent bouts of mild injury, and his slender physique. This becomes retrospective, but his inability to "put away" batters with two strikes is also indicative of a loss of "stuff." MGL's emphasis in his post on pitch speed seemed to miss the point. It isn't how hard the ball is thrown, but rather a combination of speed, movement and location. I would suggest that for all but the most exceptional of fans, the best way to evaluate those collectively is what is the result (over multiple iterations). Here, obviously, the results were poor.

Posted 5:02 p.m., November 5, 2003 (#3) - tangotiger
there is evidence that pitchers do NOT have good and bad days

I agree with Ross, and disagree with MGL.

What we should say instead is:
"the sample data is not large enough for us to show WHO is having a good day or bad day"

Obviously, everyone has good/bad days, obviously clutch exists, obivously pitchers control the outcomes on BIP, obviously chemistry affects a team.... but for us to catch this, to know this, to say that "yes, this guy has it" ,etc, etc... we can't. There's not enough data to make it statistically significant.

So, while I would ignore what we think is a good day or bad day, I do this not because it doesn't exist, but because I can't figure it out using the data.

Posted 5:04 p.m., November 5, 2003 (#4) - tangotiger
And no one can really figure it out using the data. You have to go beyond the data.

The problem is that people will always look at the result, instead of ignoring the results completely, and concentrating only on the mechanics, the delivery, the movement, the demeanour, etc.

Posted 7:16 p.m., November 5, 2003 (#5) - RossCW
Also informing my view is Pedro's history of poorer performance with higher pitch counts

I think this is a mistake. While poor performance in the 75-100 pitch range may indicate a lack of endurance, above that it may simply reflect the decisions of the manager. If Little almost always left Martinez in until he got poor results, then he is always going to have bad stats for the last pitches of the day.

Instead, I think you have to look at how often he goes well beyond 100 pitches. Presumably if he wasn't getting results he would be lifted before he got to 130 pitches.

Posted 7:48 p.m., November 5, 2003 (#6) - Arvin Hsu
FWIW, my take on the matter.

there is evidence that pitchers do NOT have good and bad days

I am guessing that MGL is referring to the fact that what seems like good & bad days to casual observers models exactly what we would expect from random noise due to a normal or a binomial distribution.

If that's the case, the issue can be restated thusly:
A is what we observe as fluctuation in performance.
A can not be distinguished from random noise B.
What do we mean by random noise B?

One interpretation is this:
B results from the binomial distribution. e.g. a machine bats a ball with a certain ability .300. This means there will be some stretches where the machine will go 0-5, and some days the machine will go 4-5.
These hot & cold streaks are strictly the result of having a probabilistic engine governing the underlying event.

Another interpretation is this:
B results from random variation, eg. like that seen in a normal distribution. We expect the batter to bat .300 on average, but he fluctuates above and below that, and it looks like he has a variance of .0004(or std-dev of .020). Thus, in 67% of samples taken, he's within +/-.20 of .300.

This latter interpretation is problematic: Every measurement in the world has an aspect of indeterminism to it, and what distinguishes the fact that some things have small variance and some things have large variance are due to causal factors underneath. These causal factors may be "good and bad days." Why do they look normal? Central Limit Theorem. Everything looks normal when taken with enough samples.

-Arvin

PS> this is my 20-second take on the matter. I haven't thought it through

Posted 7:53 p.m., November 5, 2003 (#7) - J Cross
I agree with RossCW that pitchers do have good days and bad days. Although you might not be able to tell which kind of day a pitcher is having from the results of a few batters you probably can tell by looking at velocity, location and movement. Armando Benitez seems to have days where he can throw 95 or better and days when he can't do better than 88. I'd bet that almost all pitchers can throw harder on some days than others and that this could be demonstrated by someone who kept track. Although it would be harder to prove, I'd bet that pitchers also have days when they have better movement and velocity than other days.

Posted 8:00 p.m., November 5, 2003 (#8) - J Cross
That should have been "movement and location" in the last line.

btw, I think pitchers as a rule start to lose it as they get up in pitches. Therefore, I'd need evidence that Pedro was different from other pitchers in this respect (or at least reason to believe he was different on that given day) to think that it was a good idea to leave him in.

Posted 2:21 p.m., November 6, 2003 (#9) - ColinM
Nice explanation Arvin. It makes me think about my strato days. Even Roger Clemens would sometimes give up 7 runs in 3 innings, but I doubt the strato card was missing its "stuff" that day. (Of course we all managed the game as if that was the case, I'd yank him for Bob Stanley real quick).

Anyway, I do agree with most of you that pitchers have bad days and I'm sure managers and coaches can probably recognize that even if the results might not show it.

Posted 3:26 p.m., November 6, 2003 (#10) - tangotiger (homepage)
You've all heard how Pedro doesn't have the stamina at the over 105 pitch count limit right?

Please click on the above link.

And to all you analysts who based your opinions on 100 PA: you should know better.

Posted 5:22 p.m., November 6, 2003 (#11) - Walt Davis
You've all heard how Pedro doesn't have the stamina at the over 105 pitch count limit right?

Please click on the above link.

I'm not sure what we were supposed to click through to or how much we were supposed to read. But in the linked thread there's a link to Pedro's career numbers over 105 pitches and they're outstanding.

But I don't think anyone is saying that Pedro has never been able to pitch past 105 pitches. The man used to be one of the more durable starters in the game.

And to all you analysts who based your opinions on 100 PA: you should know better.

Well, sometimes that's all you have. And while the mean may not be a particularly reliable estimate, it's still the best estimate of central tendency we have (unless you want to go for the median) regardless of sample size.

In your post on that other thread, you gave Pedro's 2001-2003 #'s after 105 pitches:

21 ip, 25 hits, 12 walks, and 26 K, with 0 HR and 9 ER (10 R).

Here are his numbers pre-2001:

148 ip, 119 hits, 37 walks, 176 K, 9 HR, 41 ER.

The hit rate and walk rate are much, much higher the last three years. I don't have batters faced or I'd do a quick test, but I suspect they are statistically significantly different. The ERA is certainly much higher (3.86 vs 2.49).

I dunno, but Pedro (Pedro!) walking 12 guys in 21 innings looks like a clear sign of trouble to me. That's a 5.1/9 walk rate for a guy with a career 2.4/9 walk rate. 37 baserunners in 21 innings is a WHIP of 1.76 ... enough to make Jose Lima blush and enough to make us think that 3.86 ERA is a bit lucky (or the result of good bullpen support). Small sample or no, that screams ineffectiveness for a pitcher of Pedro's quality. Could it be random? You betcha. Is it likely to be purely random? No.

Was it "obvious" Pedro was tired? That's a legit question. But in defense of all those folks out there who were "FULL OF CRAP", we'll note that most of them were screaming for Pedro to be taken out well before the outcome was known. It's completely unfair to accuse these folks of judging it based on the outcome.

What would they have said if he'd gotten through that inning? Well, I've been in that situation plenty of times and I can assure you that, unless the pitcher does something like blow the next 3 batters away, I say "boy, they got lucky there."

Posted 6:03 p.m., November 6, 2003 (#12) - David Smyth
The interesting thing from this thread is whether a manager should remove a pitcher based on 1) visual observation of his mechanics, velocity, etc., 2) results only, or 3) statistical info only, such as his performance after x pitches, etc.

There are advocates for each approach. If I were forced to pick one out of those three, and for a season length sample, I would probably go with #3, the statistical tendencies, properly regressed and interpreted. But the real truth is probably some mix of all three, reflecting the general truth that more info is better than less info, assuming a fixed quality of interpreter (whether mgl or Walt Davis or a typical manager). And the fact that it was an elimination game game has to be factored in, in the most appropriate way. Nobody has really discussed how that factor did/would influence their choice.

Me, I think I would have sent Pedro out for the 8th, but on a much shorter leash than G Little. First baserunner or hard hit out, he's outta there. That is, admittedly, not the most scientific way to integrate all of these concepts, but that would have likely been my way of combining them in that real-life situation.

Posted 8:16 p.m., November 6, 2003 (#13) - Tangotiger
Walt, Pedro walked .12 walks per batter in the post-105 2001-2003, and .06 walks per batter pre-2001. 2 SD over 100 PA = .05. So, it may look like alot, I'm not that impressed. As noted, his $H was .400. You've gotta figure that since even bad pitchers don't have a .400, that Pedro was probably extremely unlucky on BIP. His ERA was a direct consequence of the $H.

And, even if Pedro is actually that line I showed in 2001-2003, he's still a league average pitcher, and certainly no Jeff Weaver. Even if he was tired, he went from being superhuman to mortal. And that's the worst-case.

Posted 11:18 p.m., November 6, 2003 (#14) - RossCW
You've gotta figure that since even bad pitchers don't have a .400, that Pedro was probably extremely unlucky on BIP.

There are a lot of pitchers who have had years with BABIP over .400 - they just didn't get to pitch very many innings. To assume that a tired Pedro would pitch better is, I think, a mistake.

Posted 11:28 p.m., November 6, 2003 (#15) - Tangotiger
Why was he tired in 2001-2003 and not earlier? Injury? Here, compile a list of injured pitchers that you think might be susceptible to hitting a wall more than others.... without looking up the stats. Just give me 10 pitchers. Then, let's look to see what their $H was in 2001-2003. My guess is that they will be just slightly worse than their overall average. (i.e., if their overall average was .300, then their post-105 pitch $H would be .310). No cheating now.

Posted 1:49 a.m., November 7, 2003 (#16) - JBH
Logically, one would expect Pedro to be worse after 100 pitches.
Subjectively, he's a different pitcher after 100 pitches (fewer breaking pitches, less control, more velocity if the game is important).
Objectively, he's performed worse after 100 pitches.

At some point, don't you have to put two and two together?

Is it really likely that his post-100 pitches split is due to pure chance, the logical expectation for his performance is wrong, AND him subjectively changing his pitching style results in no lessened effectiveness?

Posted 9:25 a.m., November 7, 2003 (#17) - dlf
I understand why we would want to look at an entire career rather than a tiny slice of it. But here, there is a clear reason to look at 2001 and forward: in 2001, Pedro was injured and missed half a season. In the two years after that, he has averaged about 15% fewer IP per year than he did in the five years leading up to the injury. While still a great, great pitcher, he has "obviously" exhibited less durability within a season. I would respectfully suggest that it is equally apparent that he has shown less durability within a particular game and, thus, the 100+ pitch numbers from earlier should be given much less weight than those more recent ones.

Tango, without looking at numbers, I am hard pressed to think of very slender hard throwers at even remotely the same level of ability as Martinez. Ron Guidry and Bret Saberhagen both exhibited severe on/off seasons perhaps related to overwork, but I don't know if they showed the same in game deterioration as observed in Martinez. Perhaps Pedro's big brother Ramon. Roy Oswalt? There are many more relievers who would broadly fit into the group (Billy Wagner, Mariano Rivera, Tom Gordon) but obviously those help none at all in trying to measure post 100 pitch effectiveness.

But I wonder if you are making a fundamental error assuming that most (any?) change in BABIP is primarily luck. Yes, there is little difference between MLB pitchers in $H, but I haven't seen a study that says the same holds true at lower levels (minors, college, Japan, etc.) It could very well be that as a pitcher tires, his ability to prevent hits on balls in play compared to the general population decreases at a different rate than his ability to get Ks or prevent BBs. What evidence do you have to suggest that $H is consistent from pitch 1 to pitch 115?

Posted 9:38 a.m., November 7, 2003 (#18) - tangotiger
What evidence do you have to suggest that $H is consistent from pitch 1 to pitch 115?

Realizing that not the same pitchers make up each pool, and realizing that even great pitchers like RJ etc don't have a $H that is that distinguishable from the average Joe MLB pitcher....

Pitch $H
01-15 0.290
16-30 0.285
31-45 0.284
46-60 0.281
61-75 0.287
76-90 0.279
91-105 0.281
106-20 0.288

Posted 9:39 a.m., November 7, 2003 (#19) - tangotiger
That's 2003 AL.

Posted 9:51 a.m., November 7, 2003 (#20) - tangotiger
I took all pitchers from 1994 to 2002, and selected only those pitchers who had, at most, 100 BFP in that season. You have to figure that such pitchers probably produced pretty bad. Here are there overall totals:

ERA: 6.67 (4.51)
H/9IP: 10.9 (9.3)
HR/9IP: 1.5 (1.1)
BB/9IP: 5.1 (3.5)
K/9IP: 6.3 (6.5)

League average totals in ().

You have to admit, that's a pretty sucky pitcher, and probably representative of what you think about with a replacement-type pitcher. His ERA+ is a paltry 68. And what do you think his $H was? 0.312. The league average is .285.

If you take the league BB and divide by this group's BB, you will get BB+. That figure is 69. The HR+ is 71. The H+ is 91.

As we've found out with the Allen/Hsu model, 1 SD in $H is around .010. We expect 95% of pitchers to be within .020 of the league average. The difference between the paltry group and the league average $H is .027 (or 3 SD).

Do you think it's reasonable to think that Pedro's $H would be .100 different from the his mean when he's tired (TEN SD)? His BB+ was 50 (compared to his mean). His ERA+ (relative to his mean) was comparable to the paltry group. I think that, at most reasonable, Pedro's $H would have increased by about .030 because he was tired. .100? No way.

Posted 10:04 a.m., November 7, 2003 (#21) - Mike Emeigh(e-mail)
There are a lot of pitchers who have had years with BABIP over .400 - they just didn't get to pitch very many innings.

Exactly. I'm not sure exactly where the breakpoint is (it's at least .350, and might even be lower than that), but pitchers whose BA/BIP exceeds a certain threshold aren't *hit-unlucky* - they don't have the stuff to get major league hitters out on a consistent basis, they get hammered when the ball is put into play (allowing a high percentage of line drives), and they get pulled from the rotation very quickly most of the time when that happens.

-- MWE

Posted 10:40 a.m., November 7, 2003 (#22) - Mike Emeigh(e-mail)
There were 18 pitchers in the NL in 2003 who had a BA/BIP of .400 or higher. The most innings thrown by any one of them was 21, by Blaine Neal. As a group, they threw a total of 120 innings, 6 2/3 IP per pitcher.

The highest BA/BIP against any pitcher with 100 or more innings was .381, for Glendon Rusch. He was the only pitcher with 100 or more innings who was over .350. There were 32 other NL pitchers between .350 and .399, and those 32 threw a total of 584 1/3 innings, about 18 1/3 IP per pitcher.

-- MWE

Posted 10:52 a.m., November 7, 2003 (#23) - Jim
The good days/bad days question sound a lot like the hot hand (or cold hand) in basketball. Or streak hitting. Lot of research done on shooters and hitters as far as the predictive value of streaks, but very little on pitchers. You'd probably have to consider the in-game fatigue factor (high pitch counts). Might there something different about the mechanics of pitching (perhaps injury-related) that makes it more worthwhile to consider how a pitcher is performing that day in deciding how long to leave him in?

Posted 11:00 a.m., November 7, 2003 (#24) - tangotiger
I just want to point out that selecting pitchers based on $H, and then giving those numbers is extremely biased.

What I did, selecting pitchers based on number of PAs is also biased, but not as much. And, that bias would give a sort of upper boundary. By selecting pitchers with at most 100 PAs, we are getting:
a) pitchers who suck
b) pitchers who got hurt
c) pitchers who were unlucky, and weren't given the chance
d) pitchers who got called up, and might have a good chance next year

I think that having a pitcher with that ERA/BB/HR line was pretty much a "worst-case" scenario. That is, that's as bad as a pitcher can reasonably be expected to pitch at the MLB level. And his $H was only .030 points worse than the league average.

In conjunction with our best-guess that the true talent level of a pitcher's $H has 1 SD = .010, then I would say our best-guess of a pitcher getting tired will have his $H be worse than his normal level by .030.

Posted 1:25 p.m., November 7, 2003 (#25) - RossCW
The highest BA/BIP against any pitcher with 100 or more innings

The 100 IP standard eliminates every pitcher who was not a starter. In 2002 here are the players with BABIP over .400 who also got over 30 outs:

NAME YEARID BABIP OUTS
========================= ============
Jaret Wright 2002 0.488 55
Kane Davis 2002 0.456 42
Bret Prinz 2002 0.452 40
Doug Nickle 2002 0.442 35
Victor Santos 2002 0.440 78
Jerrod Riggan 2002 0.413 99
Mark Lukasiewicz 2002 0.410 42
Matt Anderson 2002 0.410 33
Mark Corey 2002 0.407 36
Paul Abbott 2002 0.405 79
Juan Rincon 2002 0.402 86
Sterling Hitchcock 2002 0.402 118

It would be a mistake to think that very high BABIP are limited to pitchers who never were and never will be major league pitchers.

I took all pitchers from 1994 to 2002, and selected only those pitchers who had, at most, 100 BFP in that season. You have to figure that such pitchers probably produced pretty bad

Not really. You have pitchers who got september callups, pitchers who pitched well and were hurt and a whole collection of pitchers who may have only pitched in one or two games. Its not at all clear what category makes up the bulk of the innings that create your averages. It would be reasonable to assume that those who pitched better also faced more batters.

that is that distinguishable from the average Joe MLB pitcher....

Pitch $H
01-15 0.290
16-30 0.285
31-45 0.284
46-60 0.281
61-75 0.287
76-90 0.279
91-105 0.281
106-20 0.288

I find it hard to believe those numbers - what are they based on? We either have to conclude that pitchers never need to be taken out of a game or that managers are quite skillful at removing them before they do any damage.

Posted 2:05 p.m., November 7, 2003 (#26) - J Cross

Pitch $H
01-15 0.290
16-30 0.285
31-45 0.284
46-60 0.281
61-75 0.287
76-90 0.279
91-105 0.281
106-20 0.288

or maybe pitchers just fall off in K%, BB% and HR%. Tango, do you have those numbers? btw, is the sample large enough that the differences between innings is significant? Is the first inning BIPr high because of the good hitters/speedsters who hit in that inning?

Posted 2:06 p.m., November 7, 2003 (#27) - tangotiger
In post 19, I said those numbers are from 2003 AL. In actual fact, if I show you the AVG, OBA, and SLG, you will be quite surprised. Here it is for AL, 2003

AVG OBP SLG
01-15 0.267 0.331 0.420
16-30 0.256 0.331 0.403
31-45 0.267 0.329 0.426
46-60 0.275 0.340 0.452
61-75 0.273 0.327 0.427
76-90 0.274 0.333 0.436
91-105 0.266 0.329 0.422
106-20 0.265 0.333 0.408

Realizing that the very good pitchers would dominate the 106-20 category, what you have are very good pitchers that end up pitching like a league average type pitcher. I would also say that you statement is accurate, that managers probably ARE skillful at removing pitchers.

You have to figure that such pitchers probably produced pretty bad

Not really. You have pitchers who got september callups, pitchers who pitched well and were ...

I already said that in post #24.

Please explain to me what kind of pitcher would put up the following

ERA: 6.67 (4.51)
H/9IP: 10.9 (9.3)
HR/9IP: 1.5 (1.1)
BB/9IP: 5.1 (3.5)
K/9IP: 6.3 (6.5)
$H: .312 (.285)

I'm giving you here a pitcher with an ERA of 6.67, with 5.1 walks, and K/BB ratio of around 1.2, and who gives up alot of HR. And yet, and yet, his $H is "only" .027 above the league mean.

What does this suggest to you?

Posted 2:08 p.m., November 7, 2003 (#28) - J Cross
I meant to say "first 15 pitches BIPr" not "first inning BIPr." Anyway, it seems unlikely that those splits mean anything.

Posted 2:29 p.m., November 7, 2003 (#29) - tangotiger
Here's some interesting stuff. Using ALL pitchers from 1994-2002, here's their performance lines, when:

min .20 BB / PA

H ER HR BB SO PA $H
10.28 9.28 1.46 12.43 7.37 49.71 0.310

(about 5000 PA).

Ok, so I selected guys who just performed really really bad at walks. Their ERA was pretty high, they allowed a ton of HR, struck out at a good rate, and their $H was only .310.

max .05 SO / PA
H ER HR BB SO PA $H
14.19 9.15 1.60 6.41 1.36 47.59 0.329

(about 4000 PA)

Again, truly awful numbers, lots of HR, lots of BB, very few K... and still only $H of .329.

min .08 HR / PA
H ER HR BB SO PA $H
13.94 11.15 4.79 4.94 6.40 45.88 0.308

(about 4000 PA)

Once again, a truly awful performance, with a super-high ERA, lots of walks, a ton of HR, average # of K.... and a $H of .308.

Don't you find it strange that guys who perform rather poorly in either walks, K, or HR will also perform poorly in at least one of the other areas, and yet, have a $H that is bad but not horrible?

Posted 2:36 p.m., November 7, 2003 (#30) - tangotiger
And all this $H is pretty much expected. In the time period, using my own PA calculations, the average $H was .285. We know, using the Allen/Hsu model, that the "true talent" rate of $H for a pitcher follows a distribution where 1 SD = .010. This means that we'd expect virtually every pitcher to have a "true talent rate" $H to be between .255 and .315. That essentially gives you the bounds at which the true talent rate of a pitcher will perform, with respect to preventing hits on BIP.

And, selecting pitchers based on either truly horrible BB,HR, or K rate, and the $H were between .310 and .329.

I would bet that the floor for a pitcher's $H performance, at the MLB level, is around .330.

Posted 2:42 p.m., November 7, 2003 (#31) - tangotiger
One last one.

Looking for across the board ineptness:
max .10 SO / PA
min .15 BB / PA
min .05 HR / PA

Here's your completely pathetic pitcher:
H ER HR BB SO PA $H
13.3 10.1 2.6 7.8 2.3 48.1 0.302

(about 2500 PA total)

So, you have a pitcher that gives up a ton of HR (2.5 x the league average), a pitcher that gives up a ton of walks (2.5 x the league averaage), a pitcher that can't strike anyone out (the league is 2.5 x higher), and yet.... and yet... somehow, these truly inept pitchers managed to get a .302 hits per ball in park.

Posted 9:38 p.m., November 7, 2003 (#32) - Scoriano
�After the seventh inning I told the pitching coach (Dave Wallace) to get left-hander Alan Embree and right-hander Mike Timlin ready because I was tired,� Martinez said in an interview with the Dominican press.

http://www.msnbc.com/news/990777.asp?0cv=CB20

Posted 10:14 p.m., November 7, 2003 (#33) - RossCW
"their $H was only .310."

I am not sure what "only" means in this context. That is about the same as the best hitting team in baseball.

Posted 10:29 p.m., November 7, 2003 (#34) - Tangotiger
Grrrr....

Posted 10:53 p.m., November 7, 2003 (#35) - Tangotiger
Ok, 15 minutes have elapsed now, so I'm nice and calm.

What f-ck-ng p-sses me off about someone like.... well, guess.... is that I plant a whole g-dd-mn forest, and the best comeback is about s-me f-ck-ng fern. G-dd-mn it. Why bother coming here, other than to p-ss me off, and not illuminate anything. G-dd-mn it.

Uhmmmm... sorry about that.. somehow Cartman from South Park took over.

Let me try to repeat my point, so please try to look at the broader picture.

I looked for the absolutely worst-performance in a season by a pitcher, from 1994 to 2002. I was looking for across the board ineptness:
max .10 SO / PA
min .15 BB / PA
min .05 HR / PA

Any way you slice it, if you come across that kind of performance, that pitcher pretty much sucked.

Here's what 2500 PAs of that actually did, on a per 9IP basis:
H - 13.3
ER - 10.1
HR - 2.6
BB - 7.8
SO - 2.3
$H - 0.302

Now, that is one helluva bad performance. I specifically chose pitchers who gave up tons of HR, gave up tons of BB, and couldn't strike anybody out. 2500 PAs of that. That's one pretty terrible performance. But, somehow, they managed a .302 hits / BIP, where the league was .285.

Don't you find it rather odd that I can find such a horrible set of performance, and when I look at the one variable that I did not select from, that I got a $H that was just a bit worse than average? Wouldn't you have expected a $H of somewhere in the .350 to .400 level?

But that's not at all what we found.

I maintain that at the MLB level, the true talent level for $H is probably around .330 at worst-case. And therefore, I find it virtually impossible to believe that Pedro Martinez, even if he was supposedly a tired and worn out pitcher (yet still managed to strike out 26% of his batters), could have a $H of .400 by his ability. I contend that it was almost definitely incredibly bad luck.

26 K, 12 BB and 0 HR, over 21 IP is a rather decent performance, and that's what Pedro did in pitch count 106+ from 2001-2003.

Pedro may actually have been tired, and maybe his performance dropped drastically.... but his performance was still league average at worst.

F-ck!

*****
Apologies all-round. If you read the 2nd paragraph as Cartman, maybe it'll absolve me.

Posted 5:45 p.m., November 8, 2003 (#36) - Greg Tamer(e-mail)
Ok, 15 minutes have elapsed now, so I'm nice and calm.

I would have really enjoyed reading your post if you had *not* waited 15 minutes!

Very interesting. However, isn't this somewhat biased precisely because this group you've selected give up a ton of HR and a ton of walks? I suppose this is somewhat offset by the inability to strike hitters out, but the really good hitters facing these inept pitchers are most likely either walking or crushing the ball out of the park. Thus, perhaps a poorer quality of hitters and their results against inept pitchers are being used to calculate the hits per ball in park average for this group of pitchers. For an average pitcher with regards to HR/rate, BB/rate, and K/rate, perhaps the better hitters are walking less and crushing the ball less, thus pushing the average pitcher's hits per ball in park average up towards the inept pitcher's average. Does this make any sense? Am I completely off base here? I'm not trying to argue against you, but just curious if the extreme HR and BB rates affect the hits per ball in park average. Please don't go Cartman on me -- although I do make good pies.

Posted 7:15 p.m., November 8, 2003 (#37) - Tangotiger
No, you do make a good point, as I was thinking about this myself, and was waiting for someone to bring it up.

The one interesting thing about the "tired" Pedro (post 105 pitches, 2001-2003) is that even though his K/BB did become human (26/12), he still managed to not give up a HR in 21 IP. It's very possible that one of the "quirks" in DIPS is the removal of the HR from the BIP.

It's possible that a hard hit ball against Pedro-tired would simply become a hit rather than a HR against other types of pitchers. That is, we may *expect* Pedro-tired to have a $H far higher than a schlub, simple because he gets to keep the ball in the park, even tired.

Posted 10:49 p.m., November 8, 2003 (#38) - Mike Emeigh(e-mail)
I maintain that at the MLB level, the true talent level for $H is probably around .330 at worst-case.

Probably not much different than that.

I think that what happens is that, against Pedro-tired, balls that would ordinarily be either ground balls or perhaps soft line drives become solidly-hit line drives. There are not that many line-drive home runs; one usually has to get some lift on a ball in order for it to be a home run. But line drives in play become hits *far* more often than either GBIP or FBIP. It'd be interesting to see the G/F/L breakdown against Pedro-tired vs. Pedro-normal.

-- MWE

Posted 12:19 a.m., November 9, 2003 (#39) - RossCW
Don't you find it rather odd that I can find such a horrible set of performance, and when I look at the one variable that I did not select from, that I got a $H that was just a bit worse than average? Wouldn't you have expected a $H of somewhere in the .350 to .400 level?

No, I wouldn't. You are making any number of assumptions that I think are incorrect. The first is that a pitcher with absolutely no success in any phase of the game is going to continue to be allowed to pitch. Second, is that a group of pitchers specifically chosen for their low K's, high walks and high home runs has some sort of meaning - is it really surprising that group has a high ERA? No. And third is that there is some expected relationship between poor performance on HR, SO and BB and how many hits a pitcher gives up. Lets assume we took the opposite - a .350 BABIP - would you conclude that the pitcher much walk a lot of batters. How about give up a lot of homeruns? How about strikeouts - must they be lousy at strikeouts? I don't think so and see no reason to think they are necessarily related.

Finally you suggest that .310 (or .302) isn't so bad and I point out that in this context it is "really bad" - perhaps not as extreme as the extremes you used to choose your group, but still really bad. You complain that somehow that is a minor detail - it clearly isn't. Its the difference between a team hitting like the Yankees and hitting like the Tigers.

Posted 9:33 a.m., November 9, 2003 (#40) - Tangotiger
I'm saying that .330 is about as bad as bad should get, that .400 is really quite inconceivable from a true talent persective (but obviously quite expected from a sample of 63 BIP).

I'm saying that Pedro was supremely unlucky at .400 over 63 BIP (with that being only 2 SD away from his mean), and that it's more likely that a .320 or so is really representatiove of Pedro's true talent on BIP.

Once you accept that, Pedro-tired is about a lg avg pitcher.

Posted 2:25 p.m., November 9, 2003 (#41) - Mike Emeigh(e-mail)
I'm saying that .330 is about as bad as bad should get, that .400 is really quite inconceivable from a true talent persective (but obviously quite expected from a sample of 63 BIP).

If Pedro-tired gives up more line drives as a percentage of BIP than either Pedro-normal or a league-average hurler, then it's conceivable that Pedro-tired could have a $H higher than .330.

National League pitchers, in 2003, had a $H of .290. 49% of their BIP were ground balls ($H on GB was .232), 32% were fly balls ($H on FB was .122), 19% were line drives ($H on line drives was .719). If you take an NL pitcher with 450 BIP, an average distribution of BIP, and an average rate of converting the various types of BIP into outs, convert 10% of the BIP against him into LD while keeping his GB/FB ratio on the other BIP constant and not changing the rate of conversion of BIP into outs, his $H would go from .290 to .343 just because he was being hit harder.

I think that for any tired pitcher, this is exactly what happens - more LD in play, and probably harder-hit ground balls as well, all of which are less likely to be converted into outs. One needs to look at the BIP distribution of Pedro-tired's BIP; my guess is that the LDIP% for Pedro-tired is significantly higher that Pedro-normal or most ML pitchers.

-- MWE

Posted 10:42 p.m., November 9, 2003 (#42) - Mike Emeigh(e-mail)
I took a look at the pitching lines of all NL starting pitchers in games in which they went four innings or fewer (in most of which you can safely assume that they were getting hammered), and compared them to the starting lines for that same group of pitchers in games where they went more than four innings. There were 138 NL pitchers who had at least one start of four innings or fewer, a total of 331 starts in which the average IP was just under 3 per start. In the games in which they went four innings or fewer, the starters had a $H of .433; in the games in which they went more than four innings (2010 starts, averaging just under 6 1/3 IP per start), the same starters had a $H of .278. The starters allowed line drives on 26.5% of their balls in play when they didn't make it past the fourth, as compared to 18.2% of balls in play when they did. Finally, the defense converted fewer of *all* of the ball types in play when the starters weren't getting through four innings. When the starters lasted more than four innings, the $H values were .224 on GBIP, .113 on FBIP, and .699 on LDIP. When the starters lasted four or fewer innings, the $H values were .347 on GBIP, .205 on FBIP, and .838 on LDIP. If the balls had been fielded behind the starters at the same rate in the short starts as they were in the longer starts, the $H would have been .316.

Looking at Glendon Rusch's 19 starts, in which he posted a $H of .393 and an LDIP% of 25.1%: His $H on GBIP was .301, on FBIP was .179, and on LDIP was .818 - also all well above the league average.

What appears to be happening here is that, the harder that the pitching is being hit, the more difficult it is for the fielders to field *any* ball put into play behind him. That's not really a surprising conclusion, at least to me, and it suggests that pitchers who are being smoked - for whatever reason - could easily have a very high $H.

-- MWE

Posted 11:32 p.m., November 9, 2003 (#43) - Tangotiger
Mike, there's great data to interpret in there, and I'll take a better look tomorrow.

However, most importantly, you have serious selective sampling issues. By limiting to the max 4 innings, you can guarantee that you have "unlucky" breaks on the pitcher. Phil Birnbaum did a great article on a similar issue in one of the BTNs.

For those who don't follow, if Roger Clemens gets rocked in the 1st 2 innings, and Torre has no faith in him, he'll take him out. If, at the end of the season, you pick out all of Clemens' start of max 2 innings, what whill you find? high $H, high HR, high BB, low K. It's a given.

Posted 11:33 p.m., November 9, 2003 (#44) - Tangotiger
To continue, if Torre instead had left him in after 2, what would have happened? Hard to believe, but the best guess is that in the next 2 or 3 innings, Roger Clemens would have "reverted" to his true talent rate.

So, by selectively sampling those games where Torre took him out after 2, you are not giving Clemens a chance to show his true stuff.

Posted 8:32 a.m., November 10, 2003 (#45) - Mike Emeigh(e-mail)
So, by selectively sampling those games where Torre took him out after 2, you are not giving Clemens a chance to show his true stuff.

True - although Clemens would probably get left in longer to prove himself than, say, Jeff Weaver.

But the *true stuff* isn't really the issue here, from my vantage point. I was trying to find a set of games where it was likely that the pitcher in question was being hit hard, to see what the impact on $H was likely to be - and what I see is that, when pitchers are being hammered, not only does their hard-hit ball percentage go up, but any given BIP in play is more likely to be a hit, whether it's a ground ball, fly ball, or line drive. So while his *true stuff* might be in the .320-.330 range, his actual $H is likely to be quite a bit higher than that - not because of bad luck, but because the balls he's giving up are harder to field.

-- MWE

Posted 10:01 a.m., November 10, 2003 (#46) - tangotiger
While it would be true that he's getting hit harder, and the reason, the degree of that impact is not anywhere near as much as the observation/sample would show, simply becaue of the selective sampling issue. This is why we have the regression towards the mean concept. You can pick out all the students who get a 100 on their last test, and they simply won't get 100 on their next test. Or you can pick out all the kids with an F, and they won't get an F. What happens is that the D student follows a distribution where, by luck, he will sometimes get an F and some times get a C and alot of times get a D. By picking out all the F students in 1 test, that doesn't tell you much, other that they failed, and they got alot of wrong answers (i.e., got hit hard). And they did that, in large part, by luck. Their true talent had only something to do with failing. What we're after is trying to figure out how much of Pedro's .400 was true talent Pedro-tired, and how much was just bad luck.

Since I don't have a stats degree, I'll let those who make this their vocation discuss this with authority.

Posted 11:34 a.m., November 10, 2003 (#47) - RossCW

I'm saying that .330 is about as bad as bad should get, that .400 is really quite inconceivable from a true talent persective (but obviously quite expected from a sample of 63 BIP).

Once you accept that, Pedro-tired is about a lg avg pitcher.

So you can leave him in for 200 pitches and he will still be league average? I think you are missing the point. Pitchers ability varies widely over the course of a game. There is really no reason to think that it does not vary more widely than the average variation between major league pitchers over the course of a season. There is a point at which a low a-ball pitcher will be more effective than Pedro.

For those who don't follow, if Roger Clemens gets rocked in the 1st 2 innings, and Torre has no faith in him, he'll take him out.

If Clemens has given up a couple bloop singles and a couple mistake pitches etc. Torre is not going to lift him. Its not as if the manager can't see when a pitcher is suffering from bad luck as opposed to not having good stuff.

Posted 12:02 p.m., November 10, 2003 (#48) - tangotiger
So you can leave him in for 200 pitches and he will still be league average? I think you are missing the point.

Why do you have to say stuff like that?

Cue Cartman

!@##$$%%% - !@!#@$!@%%^ - !@!$@#!#%%#$^$

Exit Cartman

I'm looking at it from a practical human viewpoint. I think we can all accept that Pedro has no problems with the first 60 pitches he throws, and his arm would fall off at 300. Obviously. So, does Pedro hit a wall and he drop to a lower plateau at some point, or is a decline on a pitch-by-pitch basis? I don't know.

But, even if Pedro does hit a wall, and drop to a lower plateau, it certainly wouldn't be at the exact same pitch count each time. Maybe he'll hit a wall at 95 or 110 or 120 or 85 or something each time. The net effect, over a period of times, is that you'll probably notice a decline at some point. (i.e., Pedro say hits a wall at 105, with a distribution centered at 105, with maybe 1 SD = 5 pitches or something... just guessing).

So, what I'm saying is that when Pedro hits that wall, he becomes a league average pitcher.

You can probably create a function that says something like:
Pedro's healthy true talent $H = .290 (or whatever it is)

For every pitch above 90, increase his $H by .002, or something like that. It doesn't necessarily have to be linear. You can make it +.002 for every pitch 90 - 120 (for Pedro), and then increase by +.005 for pitch 121 - 150, and +.010 for 151+ pitches. (Again, numbers only for illustration).

So, at pitch 105, his true talent $H might be .320, and at pitch 125, his true talent $H might be .380 (or whatever it works out to).

Pitchers ability varies widely over the course of a game.

Cue RossCW

What's your evidence for this?

Exit RossCW

No, I doubt it. My guess is that a pitcher's true talent stays extremely static during the course of a game. The variation of his performance is probably just an expected distribution around his talent that day. His day-to-day abilities I would guess would change widely (not feeling well, tight muscle, healthy outlook, feeling psyched, or whatever). In-game? Probably those with low confidence might let themselves be susceptible. I don't think this applies to many MLB pitchers. But, I have as much evidence about this as you do.

There is really no reason to think that it does not vary more widely than the average variation between major league pitchers over the course of a season. There is a point at which a low a-ball pitcher will be more effective than Pedro.

Obviously. But, my best guess is that it's not Pedro at 105. Unless you know many low a-ball pitchers who can strike out 26% of MLB batters, as Pedro post-105 has done from 2001-2003, while maintaining a better than 2:1 K/BB ratio, and not give up any HR.

Maybe Pedro at 160? I dunno.

Posted 2:14 p.m., November 10, 2003 (#49) - RossCW
Why do you have to say stuff like that?

It was a question I asked since it seems that is where the argument you are making leads. Your seem to be claiming that it isn't possible for Martinez to get so tired that he would pitch at a level less than major league average. Is that what you are saying? If not how does the "evidence" you provide for your argument support it being true for 130 pitches and not for 200?

So, does Pedro hit a wall and he drop to a lower plateau at some point, or is a decline on a pitch-by-pitch basis? I don't know

So, what I'm saying is that when Pedro hits that wall, he becomes a league average pitcher.

As you point out it isn't clear that he ever "hits a wall". If not, then I agree there is some point in his decline where he is "league average". But isn't that simply begging the question of whether he somtimes/often drops below that point when his pitch counts get high?

What's your evidence for this?

I watch baseball games. Pitchers fall apart. Pitchers struggle in the first and settle down later. Is some of that chance attributed to skill? Yes. But it doesn't appear to me that is always the case and it doesn't appear that the people who play the game think so either.

My guess is that a pitcher's true talent stays extremely static during the course of a game.

I agree. And I misused the term "ability" - I should have said performance.

Pitchers talk about the adjustments they make all the time. They find out one of their pitches doesn't have movement or they can't get it over the plate and change their approach and the pitches they use. They adjust to the umpire's strike zone. They lose something off their fastball. They "find" their curve.

Unless you know many low a-ball pitchers who can strike out 26% of MLB batters, as Pedro post-105 has done from 2001-2003,

As you point out when Pedro starts to fade is going to vary from game to game. You are including in your averages what happens between 105 and 130 pitches even when he doesn't fade until 130 pitches.

My guess is that there are quite a number of low A ball pitchers that would do just fine striking out major league hitters. Unfortunately when they weren't striking them out, they would be either walking them or getting lit up.

Posted 2:25 p.m., November 10, 2003 (#50) - tangotiger
I just love the way things are taken out of context, or misinterpreted.

it seems that is where the argument you are making leads. Your seem to be claiming that it isn't possible for Martinez to get so tired that he would pitch at a level less than major league average. Is that what you are saying?

No, and I provided a clarification to that by saying that at some point Pedro might have a true talent of .380, and his arm would drop off at some point. So, why ask if this is what I'm saying when I clearly said that he would further decline until he drops off?

And I misused the term "ability" - I should have said performance.

There's no question that there's great variability in performance. This is true in all walks of life where you have a binary (safe/out) result.

As you point out when Pedro starts to fade is going to vary from game to game. You are including in your averages what happens between 105 and 130 pitches even when he doesn't fade until 130 pitches.

Bingo! Bingo!!! This was to all those people who bring up Pedro's post-105 PAs, when they can't even say whether the true talent level he was showing at game 7 at pitch count 106 was the same, better, or worse than his average true talent level at pitch count 106 from 2001-2003.

I actually said:

Unless you know many low a-ball pitchers who can strike out 26% of MLB batters...while maintaining a better than 2:1 K/BB ratio, and not give up any HR.

So, yes, I'm sure there are some pitchers that can strike out 25% of the batters, but there's no way they'd be able to maintain a 2:1 K/BB ratio. Nuke Lalush, off the top of my head, is one of them.

Posted 12:43 p.m., November 11, 2003 (#51) - RossCW
and I provided a clarification

Yeh - after I raised the question. As I pointed out, the evidence you provided for Martinez pitching at major league average at 130 pitches applies equally to 200.

There's no question that there's great variability in performance. This is true in all walks of life where you have a binary (safe/out) result.

I said performance - not results. Martinez ability doesn't change from pitch to pitch, his performance clearly does. And in fact there is not a binary result for pitchers - there are any number of outcomes for a particular pitch.

This was to all those people who bring up Pedro's post-105 PAs,

I agree with you on that point obviously. You can't look at results and determine performance because its the manager's job to look at performance and act before the bad results. Given the sample size above 100 pitches, it doesn't take many mistakes by a manager to make the overall results look bad.

but there's no way they'd be able to maintain a 2:1 K/BB ratio

I don't think that is really true. They would just have to serve up pitches over the meat of the plate to avoid walking batters. I'm not sure that is any different than the effect of being tired on a veteran pitcher. Their slider and curve may still break just as sharply, but it no longer catches the plate.

Posted 8:25 p.m., November 11, 2003 (#52) - MGL
As I has already said, it is extrmemly unlikely that a pitcher who has the K, BB, and HR rate that Pedro has had after 105 pitches (his historical stats) AND is still throwing in the 90's with movement and any difference in control or command is not great (observation) has a true $H of anywhere near .400. It's just not possible. In order for a pitcher to have a true $H of .400 or so would HAVE TO throw a straight fastball in the 80's (or less) with little else in terms of command or offspeed pitches. That's one of the points that Tango makes and he is correct. Yes, of course it is possible for a goos major league possible to get so tired that he has the talent of an A-ball pitcher. It is likely, however, that no manager, in any situation, even Grady Little, is going to leave a pitcher in that long. It is also likely that that kind of drop-off would occur at a VERY high pitch count (I don't think that pitchers "drop off the table" - I think that it is a gradual, though not necessarily linear, decline - but I'm not sure and it doesn't really matter to this discussion) AND if Pedro were pitching with a talent anywhere near that of an A-ball pitcher, you would probably notice something severely amiss other than not throwing good 2-strike pitches.

Mike, I'm not sure at all what you were trying to do with your quick and dirty study. Even if a pitcher never had days where his talent level differed from any other day, he would have stretches where any combination of hard/soft/line drive/fly/ground balls were hit due to chance alone. And of course, all pitchers who were taken out after less than 4 innings by definition would have had an unusally high number of hard hit balls AND line drives. With all due respect, I really don't see what the point of presenting the data was.

What you want to do is to look at all pitching starts in which a pitcher gets hit hard in the first 3 innings (high percentage of line drives and high percentage of hits per GB and FB) and then look at their $H (or whatever stat you want to use to represent how "hard" they are being hit) in the next inning. That will tell you something about whether enough pitchers fluctuate significantly enough in their "stuff" (ability as opposed to performance) that it is "correct" to take out a good pitcher after he gets shelled for several innings even though you might be making a mistake (it might just have been bad luck - BTW Ross, there are 2 kinds of bad luck in this regard; one is when bloop hits fall at the right time and a bunch of runs score accordingly; the other is when a pitcher gets hit hard even though his true "talent" has not changed). My guess is that in the next inning all of these pitchers as a group will revert back to very near their normal stats (in fact, I'd bet a lot at even money that this is true). Like clutch hitting, such a test would not prove or disprove that pitchers' talent fluctutes from day to day or even from inning (or batter) to inning; it merely would evidence, one way or another, whether one could use getting "hit hard" in any given number if innings as a proxy for this talent fluctuation or whether there is not enough prevalence in this regard (either not enough pitchers who do fluctuate in talent and/or the fluctuations in talent are not that great) to be able to distinguish it from the inevitable fluctuation in performance due to the random (binomial) nature of the events. Again, my guess is that the idea that pitchers who are getting "hit hard" (or not) is significantly indicative of their true talent at the time (IOW, predictive of the future) is another of those truisms that turn out to be clearly not true. Tango and I have been debunking many such myths lately and will present some of them in due time...

Posted 3:22 a.m., November 12, 2003 (#53) - chris p
i don't know if there's any validity to this or not, but my observation of pedro during the past 2 years where he has been pedro the finesse pitcher is that he gets into trouble when he has to reach back for something "extra" or whatever. by my observation, he loses the natural movement on his pitches when he tries to throw harder. i'm curious to see a comparison of how similar pitchers (guys that rely on a fastball with good movement) tire. which starters would fit this criteria?

Posted 1:48 p.m., November 13, 2003 (#54) - RossCW
is still throwing in the 90's with movement and any difference in control or command

I guess I don't think it matters what his K/BB or HR ratio is if this part were true. How exactly do you determine whether there are any differences?

In order for a pitcher to have a true $H of .400 or so would HAVE TO throw a straight fastball in the 80's (or less) with little else in terms of command or offspeed pitches

I just don't agree with this. I think it is perfectly possible for a pitcher with 90 mph fastball to have a hard time getting major league hitters out if that is all they have and they can't locate it well within the strike zone.

The question isn't what happens on the pitches batters don't hit - its what happens on the pitches where they do. If 90% of Pedro's pitches are fine he probably isn't going to struggle a lot. But if that percentage drops to 75% he is going to be offering up at least one batting practice pitch to every batter. Put another way - when 10% of his pitches are easily hittable he probably isn't going to struggle, when 25% are easily hittable he will.

no manager, in any situation, even Grady Little, is going to leave a pitcher in that long.

I agree - but a .400 BABIP is not A ball pitcher, its AAA pitching. Some pitchers put up .400 numbers in the major leagues.

it might just have been bad luck - BTW Ross, there are 2 kinds of bad luck in this regard; one is when bloop hits fall at the right time and a bunch of runs score accordingly; the other is when a pitcher gets hit hard even though his true "talent" has not changed

And managers can tell the difference in both cases. In the latter case, the manager can tell its just bad luck when batters are hitting every mistake pitch and even a few pitches that aren't. When every third pitch is a mistake it isn't bad luck.

This will show you nothing if managers can tell the difference between a pitcher who is unlucky and one that is serving up too many mistakes. The pitchers that are unlucky will still be in the game, the others won't.

my guess is that the idea that pitchers who are getting "hit hard" (or not) is significantly indicative of their true talent at the time (IOW, predictive of the future) is another of those truisms that turn out to be clearly not true.

And my guess is that it will be clear only if you assume that you are dealing with random results rather than optimized results.

Tango and I have been debunking many such myths lately

The problem is that no statistical analysis can debunk many of these "myths" since you don't have the data to do that. You have the results and you often pretend that they reflect random actions when in fact they are a result of concious manipulation.

Posted 1:59 p.m., November 13, 2003 (#55) - RossCW
Just to be clear I am not suggesting that anyone doesn't try to eliminate the "noise" but that even when it isn't possible rather than throwing up your hands and saying "this isn't possible" you go with the data you do have. The problem is that analysis based on that data is often less reliable than the subjective observations that are being "debunked".

Posted 2:13 p.m., November 13, 2003 (#56) - tangotiger
Just so that MGL doesn't put words in my mouth, these things are *not* going to be debunked. As I've said many many many times, all these things exist. Just because we can't see it, or have enough data to say anything with statistical significance, doesn't mean it doesn't exist.

However, if we can't see it, then, chances are, neither can you.

I agree - but a .400 BABIP is not A ball pitcher, its AAA pitching. Some pitchers put up .400 numbers in the major leagues.

Ugh. They put up .400 almost by random chance. How many times do we have to go through this? Our best guess is that 1 SD = .010 $H among MLB pitchers. If pitchers give up .400, it's far far more likely it's because they have few BIP and/or really bad fielding, rather than the pitcher's probably true $H.

And .400 is A ball pitcher. AAA pitcher is probably .320. Before you ask, I don't have any evidence.

Posted 2:45 p.m., November 13, 2003 (#57) - tangotiger
$H BIP=50 BIP=500
0.270 0.063 0.020
0.280 0.063 0.020
0.290 0.064 0.020
0.300 0.065 0.020
0.310 0.065 0.021
0.320 0.066 0.021
0.330 0.066 0.021

The first column would be a pitcher's true $H. We don't know, and can't ever know, a pitcher's true $H (or true anything for that matter). Our best guess is that if a league has a .300 $H, then 1 SD = .010. The above table represents 3 SD of pitchers around the mean (i.e., 99%+ of pitchers).

The second column is what 1 SD of performance would be, over 50 BIP, given the true rate of the pitcher. Essentially, 19 times out of 20, we'd expect a pitcher to perform at a $H rate that is within +/- .130 hits / BIP. A true .300 pitcher, with 50 BIP, will have an observed $H of .170 to .430.

(In fact, we should add extra variation to account for park and fielding differences.)

The third column is with BIP=500. So, 19 times in 20, a pitcher's $H will be +/- .040 within his true rate.

So, based on the above, with at least 500 BIP, we should find maybe 1 pitcher a year at the .400 level, if even. (If of course they're allowed to pitch at that level for so long.)

And what do we find from 1994 to 2002? 7 pitchers with at least 200 BIP had a $H in the .360s. So, we probably have selective sampling issues here that would prevent a poor $H from pitching even at a bad level.

Bring it down to 50 BIP? We have 31 pitchers over 9 years with a $H of over .400. There were 1232 pitchers with between 50 and 150 BIP in that time period, or 2.5%. The average # of BIP of these 31 pitchers was 74, or 1 SD = .053. So, we'd expect 95% to be between .200 and .400. We'd expect 2.5% to be under .200, and .... 2.5% to be over .400.

(There were 41 pitchers with BIP of between 50 and 150 with a $H of under .200, just about exactly what we expected.)

Posted 5:53 p.m., November 13, 2003 (#58) - MGL
My apologies for putting words into your mouth Tango. I also think you are being a little too politically correct. :) What happened to the Cartman that I knew and loved?

Oh, and why do you bother?

Posted 12:09 a.m., November 14, 2003 (#59) - RossCW
Tango -

You apparently are trying to prove that its possible for a pitcher to have a .400+ BABIP even if their "true" ability is closer to .300. I agree - it is possible.

What evidence is there that this is the explanation for every pitcher who had a BABIP over .400? Is it surprising that there aren't very many pitchers who give up 20 hits while getting 30 outs (not counting home runs) that make it to 50 balls in play. Nor is it surprising that in 2002 there were 8 pitchers between 25 and 50 BIP who had BABIP over .400.

The problem for your statistical approach is that all your data is as hopelessly contaminated as this. The player who goes over .400 is sent down. The pitcher who gets lucky and has a BABIP of only .300 when his true ability is .400 gets a few more chances but is quickly jettisoned when he returns to form. You seem to acknowledge this "selective sampling' issue in one sentence and then ignore it in the next as if having acknowledged it somehow makes it disappear.

BTW - I'm not a statistician. But you are developing standard deviations for 50 BIP and then applying them to pitchers between 50 -150 BIP. Shouldn't you be using the average, not the minimum, of the group - i.e. 100 BIP - to determine the standard deviation?

Posted 9:47 a.m., November 14, 2003 (#60) - tangotiger
Ross said:

But you are developing standard deviations for 50 BIP and then applying them to pitchers between 50 -150 BIP. Shouldn't you be using the average, not the minimum, of the group - i.e. 100 BIP - to determine the standard deviation

You must have missed this, because I said:

There were 1232 pitchers with between 50 and 150 BIP in that time period, or 2.5%. The average # of BIP of these 31 pitchers was 74, or 1 SD = .053.