THE BOOK--Playing The Percentages In Baseball

"For now, it is fair to say only that we know nothing about how strikeout rate affects a pitcher’s aging process.”

Is this a reasonable statement?

Well he’s just saying that James’ study is faulty, and his own study is incomplete. So there is no definitive answer. I think we have an idea about the subject, but don’t know anything concretely.

“For now, it is fair to say only that we know nothing about how strikeout rate affects a pitcher’s aging process.”

Is this a reasonable statement?

No, it is not.  A finding favoring one side or another, regardless of the sample error (confidence interval, or whatever you want to call it), tells us something.  There is no magic point at which a particular finding becomes “true” or a particular side of an issue is “proven” or “not proven” (Clemens would say “misproven” I think).

I have been meaning to do a similar study to this for a long time.  I have always realized that James “study” was flawed for the exact reason that David explains and describes (quite well).  (This is one reason, BTW, why you must be careful with James’ studies and “opinions”.  With all due respect to him, he is not all that capable of doing studies like this. It should be obvious, at least since DIPS, that if you have a low K group and a high K group with the same ERA, that one was lucky and the other was unlucky and that any past or prior ERA will be higher for the low K group because of regression alone.)

BTW, David’ methodology was good.  I do that all the time in my studies (I call it “forcing” both groups to have the same FIP in this case - there are numerous ways to do the forcing - lopping off the best and worst of each group is one decent way), in order to mimic a “randomized, controlled study.”

Anyway, David is right that there needs to be more research in this area, but he has done a lot to debunk the notion I think, at least the magnitude of the notion.  I would have guessed that he would have gotten the results that he did.  I see no logical reason why a low K and a high K pitcher with the same true talent level (which is NOT what James gets when he makes the regular ERA’s equal in the two groups) will not have similar career paths.  Maybe not exactly the same, but similar.  In fact, the difference he found in future IP might be simply a manager and team bias toward high K pitchers.  IOW, let’s say that 2 pitchers are at the end of their careers or just plain suck at any point in their careers.  Might the high K pitcher be allowed a little more slack (and thus get more innings) than the low K one?

---"I see no logical reason why a low K and a high K pitcher with the same true talent level will not have similar career paths.”

Really? I think the James theory makes lots of sense, and I expect that in the end, after someone does the study ‘properly’, the hi K pitchers will still come out ahead.

And if it doesn’t turn out that way, kudos to MGL for predicting that…

MGL also wrote: “It should be obvious, at least since DIPS, that if you have a low K group and a high K group with the same ERA, that one was lucky and the other was unlucky....”

That is not obvious to me at all. Having a low or high K tells us nothing about what BABIP resulted (using basic DIPS, not DIPS 2.0). If both groups had a .300 BABIP, all it tells us is that the lo K group probably had better control, or tended to be GB pitchers.

DS, you are missing the point.  It IS obvious and it IS true!  Obviously (no pun intended) if they both had the same BABIP, they both would NOT have the same ERA or true talent!

James may have have had a point about high K pitchers not having to throw hard all the time )while soft-tossers have to throw hard just to compete) or something like that (I vaguely recall), but any numbers he cited or studies he may have done or reported to support his general assertion were fatally flawed as David G aptly points out in his article.

High K pitchers may be bigger and stronger I guess, which may correlate with longer careers, I don’t know.  So there are some reasons to think that high K pitchers might have longer careers and might not decline as rapidly as low K pitchers, but I think the overall effect is MUCH less than James and others thing/thought it is/was, as David seems to find with his admittedly preliminary results.

Fine work by David.  I agree that the advantage for high-K pitchers found by James was exaggerated.  However, it does seem to me that high-K pitchers may have somewhat more potential to improve some aspects of their game as they age, to partially offset their declining K rate.  For example, Ryan and RJ cut their walk rates dramatically in the later parts of their careers, and Schilling also improved somewhat.  In contrast, a low-K pitcher probably isn’t pitching regularly in MLB at age 26 if he hasn’t already refined his control close to the maximum his innate talent allows for.  The low-K pitchers have simply been under far more pressure over their careers to reduce BBs and HRs, meaning the hi-K guys have more potential for improvement.

So if David is out there, maybe he can tell us if he sees any difference between his low-K and hi-K pitchers in terms of their changes in BB and HR rates.

MGL #6, I’m just not getting it. Will you or someone explain it to me, like I’m a child. Maybe we’re talking about different things.

OK, if anyone is still reading this, for some reason I assumed MGL was referring to the ‘real’ ERA (true talent) instead of a ‘sample’ ERA.

Anyway, by ‘lopping off’ the 50 best pitchers in the hi K group and the worst 50 in the lo k group (43% of the sample)--I can’t quite articulate why, and mgl says it’s OK--but to me this act is very questionable. Even though it helps equalize the pitcher groups, it assumes that there is nothing ‘special’ in terms of the excluded hi FIP/lo K and Low FIP/Hi K groups.  Hopefully, someone will figure a way around it.

I am certainly willing to believe that James overestimated the magnitude of the difference because of his selective sampling issues, but I still expect that the difference exists, to a worthwhile extent.

But from a practical perspective, how important is it to figure out what a pitcher will likely do in his remaining career. Pitchers are signed for 1 to 5 seasons usually, so isn’t that pretty much the time frames that should be concentrated on?

What happens if you start from the opposite direction--you take all pitchers with a FIP between 3.00 and 3.50 and look at the high- and low-K ones. Then repeat for pitchers with a FIP between 3.50 and 4.00, 4.00 and 4.50, etc.

What happens if you start from the opposite direction--you take all pitchers with a FIP between 3.00 and 3.50 and look at the high- and low-K ones. Then repeat for pitchers with a FIP between 3.50 and 4.00, 4.00 and 4.50, etc.

***

We’ll still have a bias because the high-K pitchers will likely have more innings pitched, which would still require us to remove some pitchers from the sample.

So if David is out there, maybe he can tell us if he sees any difference between his low-K and hi-K pitchers in terms of their changes in BB and HR rates.

***

Weird, I was positive I just submitted an answer to this. Anyways, I assume that the low-K pitchers will have fewer walks and home runs given that their FIPs are equivalent.

David:  What I was wondering about is what happens to low-K and hi-K pitchers in the second part of their careers.  The hi-K pitchers are a bit better overall (and pitch many more IP).  Are there any interesting differences in terms of how the two groups age?  (my theory is that the hi-K group may show more improvement on BB rate)