now based on the corrections to the original research or not

This is un-needed. Once a player has been selected to be in the pool of MLB pitchers, he becomes part of the population of pitchers with a BABIP of .300, give or take. Even Kyle Farnsworth.

Once a pitcher has been deemed to no be MLB-caliber, he now becomes part of a different population of pitchers, someone like say a .320 BABIP.

This is the funny thing. If a pitcher is in the minors, his presumed BABIP will regress toward (MLB-equivalent) .320. But, once that same pitcher is called up, his presumed BABIP will regress toward (MLB) .300. You regress the player toward the population he was drawn from (which can comprise of far more of just MLB/non-MLB, like first-round pick, or fat, or under 30, or fastball over 95mph, etc).

This is not only true of BABIP, but of every metric around.

you're making an inference based on facts not in evidence.

My evidence is James' own, repeated testimony that he does not read the work of other sabermetricians. What better evidence do I need?

Look, I'm a huge James fan. His contributions and influence can't be questioned. It would be nice if he were someone who chose to use his considerable visibility to bring attention to lesser-knowns who have done important work in his footsteps. But if he chooses not to, that's fine. What I don't want to see happen -- as a James admirer -- if for him to start trotting out "new" work that is not better than, and maybe very similar to, work that others have already done.

Well said, Mike. But tell me: are the DIP and FIP numbers that get tossed around now based on the corrections to the original research or not?


I don't know the answer. I presume this is rhetorical and the answer is "Yes, every reasearcher that has attempted to quantify the pitcher's ability on defense has just corrected the original DIPS research." I presume you could also say that about ERA if inclined.

Nevertheless, the question I asked to RonJohn still stands. Why does it matter if the error was eventually corrected. If the debate is about " studying the components of hit prevention ability" and the position was "That isn't necessary because 'the ability didn't exist in any meaningful way'" and critiques or worse decisions were made because of "the ability didn't exist in any meaningful way" then haven't we got a real problem on the interval until the correction is made. Wouldn't the problem be worse, if the evidence for that assertion is just repeating the calculations of the model.

THIS DOESN'T HAVE TO ABOUT DIPS. That can be true of any system.

Nevertheless, the question I asked to RonJohn still stands. Why does it matter if the error was eventually corrected. If the debate is about " studying the components of hit prevention ability" and the position was "That isn't necessary because 'the ability didn't exist in any meaningful way'. and critiques or worse decisions were made because of 'the ability didn't exist in any meaningful way' then haven't we got a real problem on the interval until the correction is made. Wouldn't the problem be worse, if the evidence for that assertion is just repeating the calculations of the model.

Could you please try restating this again? Maybe I'm just stupid here, but I can't figure out what this is supposed to be saying (the lack of a closing quotation mark isn't helping).

This is not only true of BABIP, but of every metric around.

Its true of any data set that has a normal distribution and continues to have a normal distribution for the projected interval. Aren't we striving for a better decision basis.

If a pitcher has TJ surgery we know they will neither progress nor regress to the mean during their recovery period. Don't we want to be able to identify factors that would allow one to determine when increases and decreases are not only likely due to the various laws of large numbers, but when they are going to persist for a period of time?

BL/205: yes, but what's your point? That researchers don't do this?

BL/205: yes, but what's your point? That researchers don't do this?

I presume that depends on the researcher. I think you strive for better results than just relying on a normal distribution. I think mgl also strives for better results. If at any time, I think you have neglected to do these things in any of your models, I trust you found that I have told you so very politely.

Nevertheless, Emeigh's original statement was that an area of research was neglected for a long period of time because there were some researchers/commenters/analysts would vocally say it didn't matter.

My supposition is this lack of research perhaps resulted in real errors in MLB decision making that would not have been made if not for the reliance on this model before the oversights were corrected.

My assertion is this certainly lead to a lot of erroneous claims of "idiot" because someone used a different decision model (that improved upon just relying on a normal distribution of data after ignoring a component).

RE: 204

Comment was edited for the quotation marks. Regarding my question to RonJohn, see the above for a similar articulation.

IOW, correcting errors does not undo the damage created while the error was relied upon. When the problem is that the model fails to account for something, showing the calculations of the model as proof of the models efficacy is fallicious.

IOW, correcting errors does not undo the damage created while the error was relied upon.

So? Why should sabermetrics be different from any other field of study where earlier models or conventional wisdom are eventually discarded as we learn more?

We're talking baseball here, not genetics or something else that really matters. So the portentous use of "damage" seems a wee bit overblown. For a while, people relied on fielding percentage as the end-all and be-all of fielding statistics. Then it was range factor. So they were wrong. We fall, we get up, we keep going. The "damage" is more to egos that to reality.

Going back a page ... wasn't the McMeeting article written by Mike Kopf rather than by Bill himself? Two decades later, I can't remember which ones, but some of the more humorous - and less analytical - pieces in any of the Abstracts were written by Kopf.

One of my favourite things about BTF is how the ads reflect (or don't) the subject matter of a thread.

At the top I have 20 books that changed America.
At the bottom I have tips for a flat belly.
While on the side I have smiley faces.

Sounds about right to me.

This site has always been the battleground--a strangely charged territory where peoples from all over come to wage war (is there an analogy to some country in Europe that's appropriate here, Paul? You're the military historian!

Belgium, Flanders, and Artois. Mayo avec mon frites, s'il vous plait. (I've now annoyed all the Flemings, with my Walloon-ish tendencies.)

DLF--I double-checked: Bill wrote McMeeting, which did in fact appear in the 1987 edition. Kopf's material is given a by-line in the ToC. He begins in '87, becomes more prominent in 1988 and in the Baseball Books.

Guy--What works for Bill is what we get, not what we might want him to do. Gregariousness or keeping up with the "competition" is not a requirement; it's a matter of temperament. Which is why I noted that I didn't take MGL's rants that seriously. As I said, we can only wait and see. Whether he "engages" any of us may or may not be relevant to the quality of his subsequent output. If you want to get a sense of it for yourself, all you have to do is subscribe to his site.

However, I do see now that your concerns are more even-handed than they came across in the earlier post. Sorry if I over-reacted.

Paul--So BTF is basically the "Low Countries", eh? Somehow that is most appropriate!! The Balkans came to mind as well.

Srul--Agreed. Fortunately we are not living in the age of the counter-Reformation, though there are many aspects of sabermetrics's "broken social circle" that seem to operate along its tenets.

I have no idea what Don's talking about there, so I'm just going to assume he meant this.