FREE hit counter and Internet traffic statistics from freestats.com

Friday, December 21, 2007

Chat Transcript 12/21

As always, thanks to everyone who participated in today's chat. Feel free to email or comment here if you have a follow-up...

10 comments:

flyingdutchman said...

Uh yeah, Dan, thanks for the chat today, and I'm sorry to be so on your case about the Adam Everett thing. Well, I'm not sorry, actually, because I'm almost certain he is not worth what your system says he's worth. What I mean is, sorry if you weren't getting enough questions to avoid my barrage, it being a winter chat and all!

I think that efforts to obtain a clear picture of defensive value are worthwhile, and I'm sure that you can uncover some truths that the naked eye might not be able to pick up on. However, do you ever stop and say, "Wait a minute, this is telling me weird things, like that Manny Ramirez isn't that bad of a defender and that Adam Everett is 39 runs better than an average MLB shortstop"? Can't these sorts of questions sometimes clue us into the fact that something is wrong?

You know, it's like, if you're reading the bible and parsing things out, trying to understand God's plan and make sense of the information the best way that you know how, isn't it helpful to stop once in a while and ask yourself, "Wait a second, does the premise make any sense? Does this information actually say anything if the original idea is just nonsense?"

Thanks.

Chris said...

So, I've been thinking about your fielding stats and I have one question. Why don't you use park factors for infielder ratings? Doesn't the length of grass at various parks play a factor on whether an infielder is able to reach a ball? I'd imagine an infielder playing on turf gets to a lot fewer balls in his area than a similar player on natural grass.

I'd think outfield grass would have a similar effect on whether a player can get to a ball in the gap, but it seems you try and take that (along with outfield dimensions) into account using your park factors.

Would it be that difficult to come up with a park factor rating for infielders as well? I'm curious as to what the results would show.

tangotiger said...

flying dutchman: UZR says the same thing about Everett, and it's a different system from Dan.

I have my own system that I published in the Hardball Times Annual (a very easy-to-grasp system, by the way), and Everett was also in the +40 to +50 plays better than his peers.

I highly suggest you read the literature on the subject.

flyingdutchman said...

Fair enough. I'll check it out. Again.

I suspect these systems are somehow picking up on the same anomaly. I just don't know what it is. This defensive stuff is elusive. Look, I hate playing this card, but if you have played baseball at a fairly high level before, it is extremely difficult to believe that a shortstop can be that many plays better than the other best players in the world. It's ridiculous. It would be very noticeable - and don't get me wrong because Everett's skill is very noticeable - but I mean spectacularly noticeable if he were doing the things for which your systems are giving him credit. It would look like Cirque Du Soleil out there.

But I'll read everything I can about it. You're both smarter than I am, so I doubt I'll find any holes that you wouldn't have already found. Either way, I suspect something is wrong.

You're all deeply invested in your systems, and I understand that, but just keep in mind that the results do need to make some sense. Do you think Curtis Granderson is an average CF? Do you think Manny Ramirez is an average LF? Is there a difference between what you think is probably true and what your numbers say?

tangotiger said...

I don't think Granderson is average nor Manny.

It may be hard to believe that someone can be 30 to 40 plays better than an average fielder, but that's the reality of it. Ozzie Smith, Brooks Robinson, Gary Pettis, etc.

It's better to disagree on the process than on the conclusions. If we all accept the process, then you must accept the conclusions.

Dan Agonistes said...

Chris, I think you're definitely correct. The infields of parks are different in the ways that you mention and certainly if there were more artificial turf parks I probably would have created park factors right away. My guess is that it wouldn't make too much difference but it's something that's on my list of things to incorporate. And no, it wouldn't be that difficult.

Dan Agonistes said...

flyingdutchman,
I appreciate your comments in the chat and your taking the time to post them.

For the record I doubt whether Granderson is average and I "know" that Manny is not average (although I did have him below average at -2.6 runs :) Part of the problem is certainly the fact that this system does not control for the quality of the pitching staff and instead makes the underlying assumption that DIPs is correct. That is, by comparing an outfielder against a baseline for all left fielders what we're doing is assuming that a fly ball or line drive hit to our particular fielder is equally difficult to turn into an out as the average ball hit to the average fielder. But the pitching staff does differ and so it's possible that is affecting certain players.

There are other issues as I mentioned in the chat such as outfielders who are good at cutting off hits but perhaps not so good at taking hits away in the gaps.

Anyway, this kind of discussion is how we learn...

flyingdutchman said...

Thank you, Tangotiger, for the lesson on scientific method, but I never asked you what part of the process is faulty. I asked you if you think it's likely that some part of this abstract and very elusive process is faulty, given the obviously erratic conclusions. This is not a question about your process, which I said I would read up on (again), but a question about common sense (a very easy-to-grasp concept, by the way).

I think Dan might be onto something with this:

Part of the problem is certainly the fact that this system does not control for the quality of the pitching staff and instead makes the underlying assumption that DIPs is correct.

We assume that on non-homer batted balls the pitchers are subject to two things - luck and the quality of the defense. What must also be true is that every fielder, to some extent, is subject to some of the same luck to which the pitcher is subject. If Adam Everett, a great defensive shortstop, is handed an inordinate number of very fieldable ground balls, it might make a +15 or +20 play SS appear to be a +39 play shortstop. Can he maintain that level? What do you think? I'm sorry that this is not a question about the process, tangotiger, so I guess you'll have to ignore it.

Yes, this is how we learn. I guess what this comes down to is whether or not Adam Everett is fundamentally a +39-play SS or not, and that, on its face, seems like an absurd conclusion to me. Honestly, I seriously doubt that Ozzie Smith was ever a +39-play SS either.

tangotiger said...

Actually, on my blog, I have a post from a year ago that looked at it from a purely common-sense perspective. (Do a search on Baseball Guts.) And in there, the case is fairly evident that your top notch players would be in the +25 to +30 plays per season, relative to average.

tangotiger said...

In case you didn't find it, here's the "common sense" approach:
http://www.insidethebook.com/ee/index.php/site/comments/using_our_baseball_guts/

In the THT2008 Annual, I think I make a fairly convincing case that the high-end SS make some 40 or so more plays than your average SS.

Common sense, my data, MGL's data, Dan's data... all independent. All in agreement.