Some Curious Things About BABIP

[jar2574]

knock --

Someone quoted the Bill James quote that said "an awful lot of what we think of as pitching is really defense."

I'd only add that sometimes pitching is an awful lot of luck. This means that some "stud" pitcher from 1974, or 1987, or 2008, might not have been a stud if his BABIP was different. And there's no reason for the game not to allow for that possibility, in my opinion.

One example: Ricky Nolasco

His stikeout / walk ratio is worse than last years, but he's still striking out almost 3 times as many guys as he walks. The biggest reason for his demotion to AAA was just bad luck -- his BABIP was almost .400

I picked him up in my fantasy league because nobody will be that unlucky forever. But for some young guys with less talent than Ricky, a stretch of bad luck will mean they get demoted and then never get back to the bigs.

Look at Ricky's numbers. BABIP fluctuates so wildly that it turned one of last year's studs into a AAA guy for a short time:
Ricky Nolasco Pitching Statistics and History - Baseball-Reference.com

HR/9
2006: 1.3
2007: 1.3
2008: 1.2
2009: 1.3

BB/9
2006: 2.6
2007: 3.8
2008: 1.8
2009: 2.7

SO/9
2006: 6.4
2007: 4.6
2008: 7.9
2009: 7.9

BABIP
2006: .286
2007: .321
2008: .278
2009: .388

ERA
2006: 4.82
2007: 5.48
2008: 3.52
2009: 7.62

[pstrickert]

Often, IRL, pitchers do not fare as well from one start to the next. They can pitch a shutout one time out -- then get shelled the next. It may be a matter of missing their spots, pitching from behind, hanging a few pitches, lack of command of one or more pitches, etc. Is that luck? Maybe a little. More likely, though, it has to do with a pitcher's ability. The more consistently a pitcher can hit his target, get ahead in the count, get their breaking ball over, etc., the more successful he will be, the better stats he will compile, etc. The less consistently a pitcher can do those things, the less successful he will be, etc. The same holds true, IMO, when examining a pitcher's career. He may be more consistent (on the whole) one season compared to the next. Is it luck? Partly, I guess. But some pitchers cannot maintain the same consistency from year to year. It could be due to a number of factors, including plain ol' ability, experience, coaching, mental toughness, etc.

Example: Derrick Turnbow (MIL) pitched in the All-Star Game a few years ago. His career has since imploded. Last year, he got sent to the Minors. He needed to throw strikes with a pitch other than his fastball. When he did that, he excelled. When he lost that ability, he failed miserably. It wasn't only a matter of how many balls were put into play -- but how hard they were hit. Batters could sit on Turnbow's fastball and rip it. Luck had very little to do with it.

[jar2574]

Originally Posted by pstrickert

Often, IRL, pitchers do not fare as well from one start to the next. They can pitch a shutout one time out -- then get shelled the next. It may be a matter of missing their spots, pitching from behind, hanging a few pitches, lack of command of one or more pitches, etc. Is that luck? Maybe a little. More likely, though, it has to do with a pitcher's ability. The more consistently a pitcher can hit his target, get ahead in the count, get their breaking ball over, etc., the more successful he will be, the better stats he will compile, etc. The less consistently a pitcher can do those things, the less successful he will be, etc. The same holds true, IMO, when examining a pitcher's career. He may be more consistent (on the whole) one season compared to the next. Is it luck? Partly, I guess. But some pitchers cannot maintain the same consistency from year to year. It could be due to a number of factors, including plain ol' ability, experience, coaching, mental toughness, etc.

Example: Derrick Turnbow (MIL) pitched in the All-Star Game a few years ago. His career has since imploded. Last year, he got sent to the Minors. He needed to throw strikes with a pitch other than his fastball. When he did that, he excelled. When he lost that ability, he failed miserably. It wasn't only a matter of how many balls were put into play -- but how hard they were hit. Batters could sit on Turnbow's fastball and rip it. Luck had very little to do with it.

A pitcher can maintain consistent HR/9, BB/9, and SO/9 rates and still have wild fluctuations in ERA due to luck.

[knockahoma]

Well, trip got put off

I'd only add that sometimes pitching is an awful lot of luck. This means that some "stud" pitcher from 1974, or 1987, or 2008, might not have been a stud if his BABIP was different. And there's no reason for the game not to allow for that possibility, in my opinion.

But I'm finding guys who remained pretty constant in their BABIP through out their careers. Apparently, with OOTP's BABIP system, there's no way to model that reality. Look at Messersmith's career BABIP with three different teams and various defensive strengths (Note how consistent his BABIP remains even as he joins a new team and also as his career dives)

W-L ERA BABIP
1969: CAL 4-2, 2.21 .209
70 Cal 16-11 2.52 .235
71 CAL 11-10, 3.01 . 241
72 CAL 8-11 2.81 .263
73 LA 14-10, 2.70 .246
74 LA 20-6 2.59 .252
75 LA 19-14 2.29 .242
76 ATL 11-11 3.04 .252
77 ATL 5-6 4.90 .288
78 NYY 0-3 5.64 .266
79 LA 2-4 4.91 .235

In the 70s, league average BABIP was somewhere between .275-.285. Messersmith consistently had a better than average BABIP. The only thing I've yet to check is adjusted stats. Even so, look at how little Messersmith's BABIP changed when he went from top rated defensive LA to the horrible defensive Atlanta Braves. Also, look at 73-76 where his BABIP remained between .242-.252!

At any rate, in OOTP, Messersmith often comes out with a .300 BABIP, a huge number in the 70s.

That's just one example of a pitcher I see with 3 or 4 year stretches of VERY consistent BABIP numbers. They're all over the place if you look at your OOTP real stats. What's really interesting is how the BABIP's inflate from 250-.275 up to .300 in the last year or two. You can almost see the guy getting older with line drives beginning to whiz by his ear like never before in his career!

[RonCo]

Not that it's directly relevant, but OOTP can model pretty consistent BABIP numbers.

Here is an OOTP pitcher's BABIP over several years:

.298
.291
.294
.276
.299
.262 *
.264 *

* Changed teams to a better pitcher's park.

Here's another pitcher's BABIP over several years:

.309
.300
.309
.322
.309
.294
.304
.286
.240
.267

Here's another:

.284
.266
.263
.268
.280
.245


Those are just three of about the first six I looked at.

[Left-handed Badger]

Originally Posted by knockahoma

In our softball league, play continues as the gopher is considered part of the umpire crew.

Probably make a better umpire than a couple I have seen.

[knockahoma]

Hey Ron. I'm glad to see those numbers. But, I've got some that really concern me.

My concern is that BABIP is too strongly tied to a team's defense. Now, remember , I'm the guy who thinks Mark Belanger can make a real difference on defense! Putting aside any philosophical concerns (which are just matters of degree, really) about BABIP, I don't like how it's modeled in the game.

Andy Messersmith was a very consistent pitcher, from his debut with California, on to the Dodgers, and then to the Braves.

I wanted to see how Messersmith did in 74 with various teams and defenses. I did several seasons with each team. Summing up, he was pretty much his good, ol' self with LA. But, most seasons, when starting with Detroit, or San Diego, his ERA would rocket up from sub 3.00 to 3.85 and occasionally above 4.00. While his K, BB, HR numbers were usually characteristic, it was his BABIP that went out of whack.

For the record, let me say, I don't believe if Messersmith had been traded to Detroit, or San Diego in 1974 that his ERA would have shot up a full point, or point and a half, just because of the defense behind him (BABIP).

I don't see an effect that large in pitchers who have been traded. Sure, if their homers and walks sky-rocket, the ERA can go off the charts (maybe a pitcher with a penchant for giving up homers gets traded to a cozy little park and gets hammered, or like Steve Blass, he suddenly can't find the plate).

But, pitching seems to be the greater issue with historical accuracy in OOTP, and I think tying pitcher performance too strongly to defense is a problem. Of course, D does affect pitchers. But would Messersmith go from 2.65 to 4.10 just by going from LA to Detroit? It didn't happen when he went from LA to the struggling Braves. He did finish up at around 3.00, which was a high ERA for him. That's a much more realistic change, I think. Real life.

To me, that's more than an import issue.

[jbergey22]

Originally Posted by knockahoma

My concern is that BABIP is too strongly tied to a team's defense. Now, remember , I'm the guy who thinks Mark Belanger can make a real difference on defense! Putting aside any philosophical concerns (which are just matters of degree, really) about BABIP, I don't like how it's modeled in the game.

What do you mean by this?

Once the ball makes contact with the bat it is up to the defense to make the out, is it not? It really cant be both ways, can it? Either you believe it is up to the defense to make the play after ball is contacted or you believe the pitcher can control more things than walks, ks and home runs. If you choose to believe the latter you are going against what the evidence is saying.

[jar2574]

Originally Posted by knockahoma

For the record, let me say, I don't believe if Messersmith had been traded to Detroit, or San Diego in 1974 that his ERA would have shot up a full point, or point and a half, just because of the defense behind him (BABIP).

I don't know how crummy Detroit or San Diego defenses were in 1974, but why wouldn't a pitcher's ERA shoot up if he was traded to a hitter's park(Colorado) or to a team with a bad defense?

If you don't believe that is possible, then in essense, you're saying that you think BABIP is in the pitcher's control. And we're just going in circles at this point.

There are players that exhibit consistent BABIPs, but they are the exception, and the data suggest that we should not interpret that consistency as having been within the pitcher's control, except in rare cases.

Whether Messersmith is one of those cases, I have no idea. But even if he was, his BABIP would probably go up in hitter's parks and in front of bad defenses.

[jbergey22]

Originally Posted by jar2574

I don't know how crummy Detroit or San Diego defenses were in 1974, but why wouldn't a pitcher's ERA shoot up if he was traded to a hitter's park(Colorado) or to a team with a bad defense?

If you don't believe that is possible, then in essense, you're saying that you think BABIP is in the pitcher's control. And we're just going in circles at this point.

There are players that exhibit consistent BABIPs, but they are the exception, and the data suggest that we should not interpret that consistency as having been within the pitcher's control, except in rare cases.

Whether Messersmith is one of those cases, I have no idea. But even if he was, his BABIP would probably go up in hitter's parks and in front of bad defenses.

Well said. Old Tiger stadium had a short porch in right field I do know that and I believe runs were quite a bit higher there from what I remember and the Padres were pretty horrible in the 70's so I guess it doesnt seem that unusual to me.

[jar2574]

Originally Posted by knockahoma

My concern is that BABIP is too strongly tied to a team's defense. Now, remember , I'm the guy who thinks Mark Belanger can make a real difference on defense! Putting aside any philosophical concerns (which are just matters of degree, really) about BABIP, I don't like how it's modeled in the game.

Yeah, I don't understand this comment either.

How would you model BABIP, if you didn't tie it to a team's defense?

What statistical data would you rely on in designing your model?

A model cannot be built solely around Mr. Messersmith. He is but one man.

[knockahoma]

Yeah, but what kind of contact is attributed, in part, to the pitcher's abilities. Voros McCracken, back in 2001, claimed a pitcher had little, or no control over his BABIP. Since, it's been shown that knuckle-ballers, fly-ball pitchers, show some degree of control. They have an effect on the batter. Usually, it's the percentage of effect that people argue over, now.

But, as you can see, Andy Messersmith's BABIP remained very consistent, though he pitched for three teams, and went from a top-notch defense in LA to a bad one in Atlanta. I really don't mind what degree of control one wants to assign a pitcher. I'm more interested if the degree assigned aligns with what we see. For example, the Messersmith issue.

Here's what OOTP is saying:

In 1974, if you took a 20 game winner with a 2.60 ERA and traded him to San Diego, his ERA would have sky-rocketed to 3.85 - 4.20. Though Messersmith's k's, BBs and HRs all remained very comparable, the San Diego defense would have caused his ERA to bloat almost a full point and a half. His BABIP would have gone from .250-ish to .300 +.

What does this mean in terms of 1974 numbers?

It means you have an elite pitcher only because he's in LA. In 1974, by my count, there were only 16 NL starters with ERA's over 4.00 (min 20 starts). So, Messersmith is now in the bottom 20-25% of starters. Now, it's notable that FOUR of those 16 were with San Diego. But, all but one were just bad pitchers. Randy Jones was the exception. The next year, he led the NL in ERA for the Padres and finished with a 3.42 life-time.

The others were Bill Grief (4.41 lifetime), Freisleben (4.30 lifetime), Spillner (4.21 lifetime). And we can't look at those numbers with today's eyes. Those are really BAD numbers in the mid 70s, when league average was around 3.60.

Meanwhile, Andy Messersmith had a 2.86 ERA.

That's too much emphasis on defense, IMO. It doesn't reflect what most often happens in real life. Andy's ERA went to 3.00 his first season in Atlanta and his BABIP matched what it had been just two years earlier in LA when he won 20 games (.257).

In other words, I don't believe even the worst major league defense by itself (BABIP) would turn an ace like Messersmith into one of the worst pitchers in the NL.

That's my concern with OOTP's modeling.

[jbergey22]

Well if you take Mike Hampton from the late 90s and put him at Colorado would you have expected his ERA to go up by 3 runs per game?

[knockahoma]

There are players that exhibit consistent BABIPs, but they are the exception, and the data suggest that we should not interpret that consistency as having been within the pitcher's control, except in rare cases.

Jar, that's the original idea. But it's been refuted by most. Here's what appears to be common wisdom on BABIP:

I'm almost positive this was lifted from Tango Tiger. I'll double-check.

The reliability of any metric will increase as its sample size will increase. For a pitcher, seasonal BABIP is a largely unreliable measure of his skill. In order for this metric to do a good job in measuring his skill, you need several seasons worth of data.

This leads to the myth that a pitcher's skill in hit prevention is mostly the product of luck: we can't see year-to-year consistency in the metric as the noise overwhelms the signal. But, as we increase the number of years, the signal can finally match the noise.

For hitters, a seasonal BABIP is a bit more indicative in explaining his skill in getting hits: the year-to-year consistency is stronger for batters than for pitchers.

It's not that the skill is non-existent, or nearly so. It's that our ruler is insufficient to measure it accurately without years of data. Of course, this is a real problem. If a pitcher shows his stuff for three or four years, we're basically saying we can't measure that, even though it exists.

[jar2574]

knock - I don't think the bold part is the consensus view and I don't think what I said has been "refuted by most." But these issues have been discussed ad nausum in the other 2 long threads, so I'll just leave it at that. We've probably read the same studies and just come to different conclusions. We're going to end up going in circles, and we probably both have better things to do.

What I will ask again, is how you would model BABIP in OOTP? If, in real life, BABIP is so difficult to measure that our ruler "is insufficient to measure [BABIP] accurately without years of data," then how should we model BABIP in the game?

To me, something that can't be measured without 4+ years of data is not a pitcher "skill" that has a huge effect on OOTP game play. I would prefer that BABIP be modeled on defense and ball park. (knuckleballers and rare exceptions excluded, of course.)

[knockahoma]

Well if you take Mike Hampton from the late 90s and put him at Colorado would you have expected his ERA to go up by 3 runs per game?

I don't know that I would have expected it. I searched like hell for Hampton's BABIP numbers. I'm beginning to hate BABIP, cuz so few sites feature it. I did find the graph below. Hampton's consistently been under league average BABIP through his career-- except in Colorado and one year in Atlanta, I think. By the graph, I can't tell whether his BABIP was out of line, or just a bit sub-par.

Was Hampton's demise in Colorado due to his defense and luck? That's BABIP as defined by some. Balls in play destroyed Hampton in Denver? I don't think that was it, at all.

I remember back then there was a lot of talk about breaking balls working less well in the high altitude. Don't know if that's accurate, or not. But back then, the ball really jumped. Hampton's troubles were mostly that his homers doubled and tripled.

NY 15-10, 3.14 217 IP, 194 HA, 151K, 99BB, 10 HRA

The next year in Colorado:

COL 14-13, 5.41 203 IP, 236HA, 123K, 85BB, 31 HRA




Mike Hampton Graphs BABIP Pitching | FanGraphs Baseball

[knockahoma]

knock - I don't think the bold part is the consensus view and I don't think what I said has been "refuted by most." But these issues have been discussed ad nausum in the other 2 long threads, so I'll just leave it at that

I totally understand wanting to avoid the frustration. I do believe that's the current wisdom on BABIP. I'd be willing to go check the biggies for their view. I want to get it right, not argue. So, I'll probably go find some stuff from James, Tango, Tippett, etc.

As far as BABIP modeled in OOTP, I think it's just an adjustment toward the pitcher. Like I said, I'm a huge Belanger fan. I think defense matters. I just don't think it matters to the extent that it can turn an Andy Messersmith into a Dave Freisleben, and vice versa.

So, if pitcher contribution to BABIP (scale 1-10) was currently at 1, I'd like to see it move to a 2, or 3, or 4. Whatever correctly models what happens to BABIP when a pitcher moves from one team to another.

I'm continuing to search, but I haven't seen any pitchers in the 70s be traded to a team with an accompanying EXPLOSION in their BABIP. I see some that can be attributed to a poorer defense by interpretation (like going from .255 to .275). But going from .243 to .310 just because you got traded? If those kinds of changes occured in the mid-70's they were rare and probably involved a veteran in his last year or two in the league.

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

[knockahoma]

Just to be clear, I'm not out to destroy defense and luck's place in BABIP.

What is the pitcher's contribution to BABIP on a scale of 1-10?

I don't have any emotional attachment to a 1, or a 3, or a 7. I just am searching for the number, or percentage that lines up with what I see when pitchers change teams and defense. What keeps Messersmith's BABIP so consistent through the years, even while going from good defensive teams to bad? If it was luck and defense, I would expect BABIPs to explode from .250 to .300 on a regular basis. I wouldn't expect 4 years in a row of a BABIP between .247 and .257. Possible, sure. But the likeliest answer?

But at least in the 70s, I'm not seeing those kinds of BABIP explosions after trades. That suggests the pitcher is exerting a decent amount of control, OR pitchers get very lucky, very consistent on BABIP regardless of all other factors for a period of three, or four years!

The first seems more likely to me.

[jar2574]

knock -

In the example you're giving, the increase in BABIP does not need to be attributed to bad defense.

It could just be a normal BABIP flucuation, like Ricky Nolasco's from 2008-2009.

But just out of curiosity I picked the first pitcher I thought of for the 1970s.

Between 1970 and 1980 Nolan Ryan's BABIP bounced around from .233 to .298 and it bounced around everytime he got traded.

Fluctuations of .030, .040 or even .050 were common from year-to-year even while he was on the same team.

http://www.baseball-reference.com/pl...01-pitch.shtml

You're talking about a .67 fluctuation in one player. That doesn't seem implausible, especially if the defense or the park is significantly different.

Edit to fix link

[RonCo]

This is going to just go round and round and round and round and round and round ...

The games in this conversation are no longer fun to play--Knock's quotes are often cherry-picked and mis-used. The saber rules of thumb are applied to the wrong questions. Knock is astounded that movement of players from park to park can (or can not) end up with major swings, when this has been proven true often ... in fact there's a great discussion in, I think Hardball Times 2007 book about Josh Beckett's move from Florida to Boston that shows how the park itself changed his entire performance.

And it just goes round and round.

OOTP models the game of baseball overall pretty well these days. I'm even more capable of saying this after my recent work with defense. Despite some holes and questions, I'm overall quite impressed with Markus's work there.

The bottom line is that if you want to make certain that every OOTP player behaves to within very close parameters to their real life historical counterparts regardless of all parameters, then the game will have to be more constrained than it is...or than it should be if OOTP is going to model baseball as we are growing to know it. In other words, you have to take the "luck" out in the areas where variance is very large (luck being defined as the wide variance of a player's performance around their actual skill, not that there is no skill to begin with). You would also have to make absolutely certain that the root ratings are correct.

If this is what Knock is saying, then I agree at least to an extent. I've always said that the attempt to be a historical, modern real, and fictional environment is causing way more difficulty than it's worth.

However, the fact that OOTP can't model the "luck of the draw" of any particular year does not invalidate the model. You can find Messersmith models in OOTP, for example...but it doesn't guarantee that Messersmith is going to be one of them.