THE BOOK--Playing The Percentages In Baseball

My guess would be related to “luck-based” factors in our equations.

We know, right now, that many things (clutch, BABIP, etc.) appear to be entirely luck-based, and out of the control of players.  And in terms of making equations and models, in some cases we ignore such factors for that reason.

But my guess is, we get better at identifying the degree to which these things are controllable, and get better about incorporating this knowledge into descriptions of performance as well as projections.

I also think we’ll get better at identifying the portion of defense that’s due to positioning/alignment of players prior to the pitch, and the portion of defense that’s due to players moving quickly to reach baseballs and making good throws to record outs.

As Tango said, we’re frank with the fact that we don’t know this stuff yet.  But that also leads us to treat it as a “0” in calculations.  So in a sense, I think these will be the types of things we “correct” in the future.

As Tango says, what things will turn out to be “wrong” is proportional to the current level of uncertainty about them.  Now, that being said, there is also a level of uncertainty around the level of uncertainty. IOW, there are some things that we feel, say, 90% certain about, and we are wrong about that certainty (we should be only 60% certain). Of course that really makes no sense.  All it means is that we should not have a 100% level of certainty about anything, even, for example, the value of the HR (and the other offensive events).

What will happen is that if we compiled a list of all the things we were 99% certain about, in 20 years (or whenever) 1 out of a 100 will turn out to be wrong. Of course, that number, 99%, is not an unbiased estimate.  It could be wrong.  It could be that we are universally over or under-optimistic about our certainties.

Anyway, I think it is a reasonable and interesting question that the reader asked - basically what things are we not that certain about right now?

One also has to make a distinction between things that we think are true but we are not nearly 99% certain of, like hitters peak at age 27 or so, things that we know we cannot measure extremely well, like defense, and things that we simply know little about, like the effect of workload on future pitcher performance and chance of injury.

I think Tango’s being a bit hard on the questioner.  He asked, what are the things that sabermetricians *believe* now that they won’t in 20 years.  That doesn’t mean what has been PROVEN now, but just what is BELIEVED now.

And, of course, it’s a very hard question to answer, because the stuff we believe now is the stuff we believe now.  My views on certain aspects of public policy are a lot different than they were 25 years ago, but there’s no way I could have predicted how they’d change back then.

20 years from now, we’ll probably look back at some of the (sabermetric or political) things we think now, and say, “holy crap, I can’t believe I thought that.”

I agree I am a being a bit hard.  From my point of view, a saberist can only believe as much as the data can show.  The rest of whatever he thinks is not a belief, just an untested theory.

So, yeah, I’m mincing words.  But that’s what I do!

To me, the “poster child” for the, “I cannot believe that we believed that” is the sac bunt…

MGL: excellent one.  I think we were very shortsighted about it.

"If we’ve already proven something right, how can we proven wrong later on?  That’s why it’s a proof.”

Great circular logic there, starting with a false premise.

A theory can not be proven, even one which is backed up by data and statistics.

At the risk of echoing earlier comments,
the biggest uncertainty with sabermetric stats at this time is the assumption that what happens on average has any relation to what is happening with a given team or player at a given moment in time.  Games are not played by league average teams with league average players and league average batting orders.

The other uncertain assumption is that just because you can not find evidence of clutch hitting, or unclutch pitching, or that catchers pitching calling can improve pitching, etc, you assume it does not exist, going against what scouts and coaches believe to be true.

Your assumptions may be true, but they are not proven, and any belief that they are may be more a matter of faith.

The greatest obstacle to the truth is the illusion one already knows it. 

As MGL said, there is uncertainty, and the degree of uncertainty is uncertain.  That’s is not just true in sabermetrics, but in almost all science where the hypothesis can not be tested in a laboratory or controlled environment. 

You say Player A created 10 more runs than Player B, in theory, but there is no way to prove it, or even disprove it, so it’s not even a theory, just a hypothesis.

Jason Bays 2009 season was valued at 16 million until last week, then with the click of a mouse, it is valued at 22 million, slightly more than Ryan Howard, who coincidentally or not, just received a contract worth much more than Bays by about the same percentage on an AAV basis.

My guess is that 10-20 years from now we will understand how pitchers induce weak contact.  That’s not to say we will prove DIPS wrong (we won’t), but those who interpret DIPS to mean that pitchers have basically very little or no control over batted ball outcomes will be shown to be wrong.

pft: thanks for completely ignoring everything else I said, and then making your point that is consistent with what I said.

#8

Since we know that the handedness of the pitcher, their K/BB/GB/FB rates, location of pitches, count on pitches, type of pitches thrown, and velocity all have at least a tiny bit of impact on pitcher BABIP, don’t we already know a big chunk of what we need to?  We know that pitchers have a true talent range around .02 on BABIP, and we know how many BIP it takes to reach r=.5, etc. for BABIP. It seems to me that we have a fairly solid handle on what is going on here.  There aren’t too many individuals who still claim that pitchers have no BABIP ability or have no ability whatsoever to induce weak contact.  If they do, they need to hang out here more often.

Matthew - A low BABIP is not equivalent to inducing weak contact.  When he are privy to Hit f/x data we will be able to describe weak contact accurately and do the kind of research that Mike is talking about.

From 1988, are any of these wrong? :

1. Minor-league batting statistics will predict major league batting performance with essentially the same reliability as previous major-league statistics.

2. Talent in baseball is not normally distributed. It is a pyramid. For every player who is 10 percent above than the average player, there are probably 20 players who are 10 pecent below average.

3. What a player hits in one ballpark may be radically different from what he would hit in another.

4. Ballplayers, as a group, reach their peak value much earlier and decline much more rapidly than people believe.

5. Players taken in the June draft coming out of college (or with at least two years of college) perform dramatically better than players drafted out of high school.

6. The chance of getting a good player with a high draft pick is substantial enough that is clearly a disastrous strategy to give up a first-round draft pick to sign a player like Rick Dempsey, Pete Falcone, or Bill Stein.

7. A power pitcher has a dramatically higher expectation for future wins than does a finesse pitcher of the same age and ability.

8. Single-season won-lost records have almost no value as an indicator of a pitcher’s contribution to a team.

9. The largest variable determining how many runs a team will score is how many times they get their leadoff man on.

10. Any one of the following:

•A great deal of what is perceived as being pitching is in fact defense.

•True shortages of talent almost never occur at the left end of the defensive spectrum.

•Rightward shifts along the defensive spectrum almost never work.

•Our idea of what makes a team good on artificial turf is not supported by any research.

•When a team improves sharply one season, they will almost always decline in the next.

•The platoon differential is real and virtually universal.

Those are from Bill James 20 years ago.  Other than some rather hard language ("same", or “almost no”, etc), the sentiment is pretty much accurate (if not precise).

What’s your point?

The original question was something like what things do we believe now that we won’t in 20 years.

So, I looked to see what things *we* believed 20 years ago as point of reference.

#11 -

Yes, I know that weak contact and BABIP are not equivalent, but Mike said “That’s not to say we will prove DIPS wrong (we won’t), but those who interpret DIPS to mean that pitchers have basically very little or no control over batted ball outcomes will be shown to be wrong.” This is much more overarching than questioning “weak contact” only and is exactly what my post was referring too.

Do you really think I would be able to list so many of the factors involved with BABIP reduction and believe that weak contact and BABIP are equivalent at the same time?

dq: good point ... of the biggest things “we” (which is mostly Bill James, but still) thought 22 years ago, we still believe them.

What about some of the smaller things?  MGL brings up the sac bunt.  Any others?  What did we believe in 1988 that we’re a bit shamefaced about now?

The primary finding about DIPS is a statistical one about what we can say about pitcher BABIP skill in a given sample size.

People have taken that finding to mean that pitchers have very limited control over and ability to induce weak contact.  I don’t believe that’s a correct interpretation of DIPS.

What Matthew lists in #10 are a lot of the factors that go into weak contact-inducing skill.  Probably not all of them.  One of the reasons that pitchers can’t turn weak-contact-inducing skill into BABIP skill over the long haul is that they can’t always throw in the situations and with the pitch types and locations that favor weak contact.  So over a large sample we can’t see that skill demonstrated.  As our data and our understanding gets better, we’ll be able to identify that skill in the smaller subsets where its effect is much larger than .020 points, and we’ll be able to take those building blocks and build them back up into a complete picture of the skill for a given pitcher.

But what about Maddux?  He doesn’t have the handedness benefit, he throws groundballs which are supposed to drive up BABIP, he doesn’t have the high-K correlation going, yet he is still almost 2 SD away from his mates in terms of preventing hits on BIP. Glavine is 1.5 SD away from his teammates in terms of preventing hits on BIP without the Ks, FB,s and most of the other BABIP reducing indicators as well.  It seems like Glavine throwing a larger % of change-ups (lower BABIP) low and away(lower BABIP) and preventing pulled balls may be the BABIP reduction factor there.  But is Maddux really creating weak contact or is there something else going on? Are there other guys who have opposite of the BABIP reducing indicators that have great BABIP numbers compared to mates or are Maddux and Glavine unique?  Like Moyer and Zito have the FB thing covered despite lack of Ks.

What we may be wrong about (in order of probability):

Players steal more than they should.

Starting pitchers are overpaid.

Age 27 is peak. (More like plateau starting at 23)

MLB teams undervalue fielding.

Also:  Derek jeter became a decent fielder in his 30s, and Leo Mazzone reduced pitchers’ ERA by .50.