THE BOOK--Playing The Percentages In Baseball

The overwhelming theme of the questions seems to be “Give me a hard and fast rule about __________ please.” and your answer is “It depends; it should be a sliding scale.”

This in fact is what a large portion of my workday consists of.  I might begin including a link to this page in my explanations.

Good stuff. Since DIPS and BABIP were such a hot topic over there, I have a follow-up that’s always bugged me a bit. Why use BABIP when BA itself doesn’t tell us much about how good a hitter is? Wouldn’t WOBABIP tell a more accurate story?

Mcsnide/2, batting average tells us a lot about how good a hitter is.  It doesn’t tell the whole story, of course.

What’s the question you are trying to answer?  You always need to identify that before you start talking about which tool is best for the job.

If the question is “Which hitter contributes the most offensive value?” then BABIP is definitely not the best tool.

I agree with Mike.  Tell us your question first, and forget about “the stats”.  Frame the question in english.

OK, let me try again.

To Mike’s point, sure, BA tells us something, but I’ll give you Ichiro’s .330 career average if you’ll give me Pujols’ .329. Even when we throw HRs out, Pujols is still better on balls in play. I guess my question is why this difference wouldn’t translate to pitchers as well?

I get the point that BABIP at this level is largely out of the pitcher’s control, but do we also know that he can’t control batted ball outcomes in other ways? Tango said at BPro,

We are presuming that a pitcher doesn’t have a disproportionate number of warning track doubles and infield flies when we look at seasonal lines.

Boiled down to its simplest form, my question is WHY are we presuming that?

Your questions are not of the “I need to do this… what metric can help me here” variety.

Instead, they are focused on the metric itself, and trying to figure their implications.

I was asking for you to do the former, not the latter.

Anyway, since you asked:

I guess my question is why this difference wouldn’t translate to pitchers as well?

You are asking why do we not look at SLG on BIP for pitchers, if we do it for hitters.  Is that correct?

The short of it is that the run value of a BIP, be it from GB or FB pitchers, are virtually identical.  That there’s really not much distinguishing them.  This is unlike hitters.

And the reason is for the reason I noted: the pitchers who suck at BIP are not in MLB.

***

Boiled down to its simplest form, my question is WHY are we presuming that?

See above.

If we ONLY look at a pitcher’s K and BB rates, and ignore EVERYTHING else (including HR), we ALREADY are most of the way there.

Yeah, sorry, I wasn’t trying to figure out what metric to use for something; I’d just always wondered if non-HR balls in play were excluded from DIPS because they’re largely meaningless or simply because they’re so difficult to quantify accurately.

Anyway, your interpretation (SLG for pitchers) of my question was spot on and I appreciate the answer.

So I have a very similar question to mcsnide that I’ll try to frame so its understandable. It is about evaluating pitchers and hitters. So the best stat (or one of the better stats) for evaluating hitters is wOBA. So why not use that stat but flip it for pitchers (say wOBA allowed)? So instead of saying this pitcher had a BAPIP of x, why not say he had a wOBA allowed of y? Wouldn’t that tell you more about the pitcher, especially in relation to how hard batters are hitting the ball off him than say BAPIP?

On a further note, could you take a pitchers wOBA allowed and use that instead of ERA or FIP in describing how good / bad a pitcher (ie. if you converted a pitchers wOBA into runs allowed similar to how wOBA is converted to runs for hitters)? Would that metric be any good? That way you could use the same metric to evaluate pitchers and hitters.

The reason I ask is that in cricket, essentially the same stat (batting average / bowling average) are used to describe how good a batter or bowler is. Why not use the same for baseball? Hope that makes sense.

Someone brought up basketball in a discussion of hitter vs pitcher BABIP a while ago and I really like the analogy. The hitter is like the shooter. He swings at, or takes shots based on, what he’s given. Pitchers determine where the ball is and how it gets there. Defenders determine where the shots come from and how they are released. It’s obvious that the shooter and defender in basketball play different roles and can affect the game differently. Yes, it’s asymmetrical, but we shouldn’t be surprised that the hitter and pitcher affect the game differently. The matchup itself isn’t symmetrical.

I am afraid that “evaluating pitchers” or “evaluating batters” is not a specific enough way to couch a question.  There are 10 different (with 10 different answers) interpretations or meanings of “evaluate.”

There was a similar question about base running linear weights (UBR) on FG today.  It could have been about any metric.  The question was, “How long does it take for a player’s UBR to be conclusive,” or something like that.  I was like, “Define conclusive.  A million games?  An infinite number of games.”

How can I answer that question when I have no idea what the person means by “conclusive.” Or even conclusive about what?

Questions have to be asked very, very specifically in order to be answered correctly.

This applies to a lot more important things than baseball…

wOBA describes the performance of the player involved.

It applies equally to pitchers and batters based on that specific definition.

However, the relevancy of the metric to the identity of the player is higher for hitters than pitchers.  That’s because there is more noise in the metric for pitchers than batters.

If you intend to mean a player’s true talent level, then each component will be regressed a different amount, and that regression will be dependent on the number of trials.

All that said, my go to stat if I have at least 2000 PA is wOBA for pitchers.

Yaz/8 - in creating the projections at the Hardball Times, I do use wOBA allowed by pitchers, and convert that to an expected ERA.

One thing this does not capture is sequencing. That may be good or bad. Some pitchers may pitch differently than the standard split between no runners/runners on (better or worse) and that is not captured. But also an ERA can be the result of bad luck, allowing a couple three run homers.

Brian

Thanks, I was not aware of that. I am new to reading up and finding information on these stats.

MGL
You’re right, ‘evaluate’ is relative term. I was thinking in terms of relative amount so would wOBA allowed by a pitcher tell us more about the pitcher’s talent than BAPIP? And if so, why even mention a pitcher’s BAPIP? I guess I really don’t like BAPIP as a stat because it seems very empty. You wouldn’t use batting average to evaluate how good a hitter is, so why would you look at BAPIP by a pitcher when there are better measures out there.

Right, BABIP for a pitcher is almost useless other than describing what happened on BIP.

FIP or something like that is best for relatively small samples and something like RA is best for really large samples.  wOBA against, or ERC (component ERA), is also good for large and medium samples.  wOBA or ERC is perfect for any size sample if you regress each component appropriately first…

And part 2:

http://www.baseballprospectus.com/article.php?articleid=14037