This could be interesting, but I don't see how BABIP is a serious factor, really. It's well known how 0-2 and 3-0 affect the plate appearance's outcome, BABIP aside.

The trick is for Bannister to figure out how to get hitters to 0-2 (suggestion: be Randy Johnson ca. 2000).

That said, I would be mildly interested to know if Jim Thome is affected at all by what is described. That dude swings so hard every time he swings, I wonder if the count makes any difference in his BABIP.

Brian Bannister needs to get out of his parents' basement and watch a game once in a while.

This is actually interesting. It may just be a fancy variation of the old truism about gettng ahead of hitters, but it doesn't hurt to dig a little deeper. It's telling that this is coming from a guy like Bannister who is univerally regarded as having mediocre stuff yet still managed to have success last year.

I really don't get this as some sort of revelation.

The spreadsheet cliche is true, if we're actually surprised that hitters don't make good contact in pitchers counts.

There's nothing wrong with verifying the hoary old saws, but to express it as if it's something new, or something that we should be surprised at seems ignorant of what actually happens on the field.

There have been a lot of new and interesting things coming out of the stat community over the last year or so, all of which are new and interesting because of the new types of data involved, but I'm not sure that too many of them are capturing anything that "baseball men" (whoever the hell they are) didn't already know.

To me the interesting implication here is not the effect on our perception of pitchers and hitters, but rather on the defense. If you have a defense whose pitching staff is constantly falling behind batters, your defensive efficiency is going to take a hit. Since it's a pretty reasonable assumption that fielders don't greatly influence the count at which a ball is put into play,* perhaps this is something worth adjusting for when rating defenses.

* It's probably true that pitchers, having a good idea of the quality of the defense behind them, may alter their approach to a batter by nibbling more when in front of a bad defense and attacking more in front of a good one.

This isn't a huge revelation, granted. "Get ahead in the count." However, it does call into question the central assumption of DIPS theory. DIPS assumes that the pitcher is a hapless victim of his own luck once the bat meets the ball. This says that count matters. Since pitchers pitch differently in different counts, perhaps pitch matters. Suddenly, the pitcher matters. The effect size might not be great, but even finding some signal in a lot of noise might be helpful.

Is there any site that gives BABIP info for minor leaguers? I know fangraphs does for a period for players that were in the minors recently (like last two years) but anywhere else that gives more detailed data?

#6: FirstInning does.

DIPS assumes that the pitcher is a hapless victim of his own luck once the bat meets the ball.

No, no it doesn't.

Granted, that's what all the controversy was about when Voros' article came out. Granted, I'm not sure even Voros understood what his "theory" was about.

DIPS Theory: After controlling for HR, BB and K rates, BABIP does not help you predict next season's ERA.

There is nothing in that statement which says that BABIP doesn't vary across pitchers. What it says is that after controlling for variation in HR, BB, and K rates (which are correlated with BABIP), the remaining variation in BABIP is not predictive of ERA.

I know of only one study that actually tested that theory, by JC Bradbury, and he found it held. But it wouldn't surprise me if others have looked at it by now and found different results.

Now, there were a lot of limitations to that study (mainly the sample) and MWE always points out relievers ... and indeed they may need to be adjusted for.

As to Bannister's notion, folks here don't seem to be quite interpreting that correctly either. What he is hypothesizing is not "hitters don't hit well behind in the count" (which indeed everyone knows) but rather "differences in pitchers' BABIP are explained (to some degree) by differences in pitchers' ability to get ahead in the count." He's leveraging on the truism that hitters don't hit well with two strikes not thinking it's some brilliant revelation. What he says is perfectly sensible (though doesn't mean he's right).

I'll simply note that an approximation of that theory would be that K/BB ratios explain differences in BABIP ... which would be consistent with BABIP not predicting ERA after controlling for K and BB rates (and HR).

I'll also make my traditional point that BABIP isn't what anybody should care about anyway. SLGIP or maybe OPSIP is what should matter. We know that flyball pitchers have lower BABIPs but this is counter-acted (at least in part) by them giving up more doubles and triples (and HRs of course).

Craig, good site. Do you know why some players have a star next to them? I couldn't figure it out.

This says that count matters. Since pitchers pitch differently in different counts, perhaps pitch matters. Suddenly, the pitcher matters.

It only says something about the pitcher-hitter dynamic. Personally, to me it suggests the batters swing defensively on 0-2 and swing really hard on 3-0.

What he says is perfectly sensible (though doesn't mean he's right).

It's not that it's not sensible. It's that it's not helpful to pitchers. Real help would be telling him how to get to 0-2.

The most useful thing here is measuring the size of difference. How much better are the odds of getting an out on a ball in play if it is 1-0 vs 0-1? Stuff like that has potential to be of value.

First thing that is needed though is to put in a weight for pitcher quality as a variable. After all, one would expect Santana to have more 0-1 and 0-2 counts than, say, your average wild young kid up for the first time. DIPS suggests there shouldn't be a difference, but what if there is? That could introduce a large bias.

Another related thing is to see if, when you factor in counts, does DIPS hold better or worse? Greg Maddux is an outlier on DIPS iirc and he is known to get more 0-1 and 0-2 counts than the average pitcher. Still, the biggest anti-DIPS theory guys are knuckleballers who one would think fall behind more than the average pitcher.

Another piece of the puzzle. Good piece.

Three cheers for Brian Bannister! I'd be pretty impressed if Tango or MGL had noticed that detail about DIPS. If the pitching thing doesn't work out, Brian is welcome to jaw endlessly about sabermetrics with me anytime he pleases.

Two things. The ahead and behind in the count stuff was mentioned in a followup on BP less than a month after the original article:

Pitching and Defense Followup

Also the great Sean Forman has made this discussion easy by listing BABIP in his splits on Baseball Reference:

2007 Splits Baseball Reference

They range from .290 (0-2) to .312 (0-0) (with the problematic exception of 3-0) and considering even the very best in the world at getting ahead of hitters gives up substantially more balls in play on 0-0 counts than any other, there's precious little advantage to be had in the stat there.

I'm betting Maddux's glove has saved more hits on balls in play than his getting ahead in counts ever did.

On K/BB ratio, I've found that strikeouts might actually be the number one determinant (before we start talking about things like line drive rate and knuckleballers) in the traditional stats in terms of lower HBIP rates. When guys talk about "better pitchers having lower HBIP rates" what they're mostly talking about is better strikeout pitchers having better rates.

However I've found no such link in the walks category, and in fact any relationship that may exist seems to indicate that "more walks = fewer hits on balls in play." That's very small if it exists at all.

considering even the very best in the world at getting ahead of hitters gives up substantially more balls in play on 0-0 counts than any other, there's precious little advantage to be had in the stat there.

This simply is not true. There were 6 pitchers in 2004-2006 that gave up more than 75 Balls in play at 0-0 and even more on 0-1. And there were quite a few more where the difference was less than 10%.

They range from .290 (0-2) to .312 (0-0) (with the problematic exception of 3-0) and considering even the very best in the world at getting ahead of hitters gives up substantially more balls in play on 0-0 counts than any other, there's precious little advantage to be had in the stat there.

These differences for the league as a whole may seem small but they are still significant. But individual pitchers can have much larger differences which they repeat year after year and are not the result of a difference in batter ability as Voros suggests.