Peter's been busy

about 3 years ago APV 21 comments 2 recs |

Story-email Email Printer Print

Comments

Awesome

by Roger Dorn on Dec 2, 2008 12:15 PM EST reply actions

Nice work, PB. Only thing I don’t like is seeing further evidence that we should expect Choo to come back to earth.

by cleveland teamer on Dec 2, 2008 12:33 PM EST reply actions

no mention of the effects of BABIP on #2 ceiling starters, specifically.

by Brick. on Dec 2, 2008 12:34 PM EST reply actions

That’s some great work. Beltre, please.

Helium Watch: Chuck Lofgren, OF

by jhon on Dec 2, 2008 12:38 PM EST reply actions

BUY LOW!

2008 Omar Vizquel .239 .277 -.038

by Brick. on Dec 2, 2008 1:12 PM EST reply actions

Well, there’s another 3B name in there — Willie Aybar. I can’t say I was overwhelmed by a quick look at his numbers (including the minor league numbers). But, Peter B. says he should play every day. Would Tampa trade him, since Longoria’s blocking him?

by peter m on Dec 2, 2008 1:22 PM EST reply actions

Here’s a link to discussion of the article

by APV on Dec 2, 2008 3:49 PM EST reply actions

as great as all of this looks, it certainly doesn’t work every player. Just because Beltre had a babip of .279, doesn’t mean he’s due for an improvement in that area. Beltre has a career babip of .292 and hasn’t been over .300 since he came to the AL. In fact, during his monster year in 04, he only sported a .328 babip. There’s obviously a chance he’s a little more lucky next year, but overall his babip numbers have been very consistent.

by JP_Frost on Dec 2, 2008 5:16 PM EST reply actions

It actually does work on every player, but there’s still more variance than correlation, and “the dice have no memory.”

It wouldn’t be that weird to be a little below average four years in a row. I mean, there’s 600 players, you’re going to have a few dozen where that happens.

by Jay on Dec 2, 2008 7:05 PM EST up reply actions

what i was trying to say is that Beltre never really had a high babip in any of his seasons and aside from 04, he’s been very steady in terms of production. That’s why I don’t see how his babip should go over .300 for him next year since he’s only done that 3 times in his career. It doesn’t seem like his babip is low due to bad luck, but maybe his overall approach is causing that.

anyway, it was a very interesting article.

by JP_Frost on Dec 2, 2008 7:45 PM EST up reply actions

Vazquez (no, not the ex-CWS choker) is the most unlucky?! I guess that means he would have hit .900 against us with luck.

Travis Hafner is overrated. Clarity is underrated. David Dellucci is David Dellucci.

by westbrook on Dec 2, 2008 8:57 PM EST reply actions

nevermind. I shoulda read the whole title first.

Travis Hafner is overrated. Clarity is underrated. David Dellucci is David Dellucci.

by westbrook on Dec 2, 2008 8:58 PM EST up reply actions

Very nice. This is a really interesting topic.

One thing I notice which they don’t explicitly point out in the article is that, of the 28 players listed with “extreme” over/underperformance in 2007, exactly half deviate in the same direction from their xBABIP in 2008, and exactly half deviate in the opposite direction, which is what you would expect if the remaining variance in BABIP is almost entirely noise. This is especially impressive since the 28 players selected were the most extreme examples — if there were additional factors hiding that they didn’t take into account, you would expect that they would show up in these players if anywhere, leading to a greater than 50% proportion who stayed on the same side of their xBABIP.

In other words, the model does a good job of sucking all the memory out of the dice.

I have lots of questions, but I’ll go read the discussion that already happened first.

by Logodaedalus on Dec 2, 2008 11:19 PM EST reply actions

I congratulate PB and associates on fine scientific effort. We now have a model that appears significantly better than previous models. There may be future models that surpass this new one. What we called noise yesterday, now partially appears to be explainable. Whose to say how much of the remaining noise will be explainable in the future?
PB would be the first to acknowledge that we have only scratched the surface of perfecting these models. There are elements of performance that we don’t even bother to measure today.
As for sucking ALL the memory out of the dice, we probably agree it’s an oversimplification.

by elsandito on Dec 3, 2008 11:20 AM EST up reply actions

Yeah, of course there will be models in the future with higher R^2 values. The ultimate criterion for pulling out all player-stable factors would be a more detailed analysis of the leftover deviations from prediction than simply looking at positive vs. negative. But looking at sign is certainly a good first pass.

We also shouldn’t expect that we’ll ever have a model of any predictive value that explains a huge amount of the variance (I mean, really explaining a third of it is already pretty huge). For something like BABIP, there is bound to be a big influence of factors that aren’t stable, individual player factors.

by Logodaedalus on Dec 3, 2008 12:33 PM EST up reply actions

Agree with your second paragraph

by Roger Dorn on Dec 3, 2008 1:04 PM EST up reply actions

“…sucking all the memory out of the dice.”

This phrase makes me think of the parasitic-leader-baby in Total Recall.

“Open your mind”

Helium Watch: Chuck Lofgren, OF

by jhon on Dec 3, 2008 1:33 PM EST up reply actions

by hans on Dec 4, 2008 12:59 PM EST up reply actions

Damn it, Cohagen …

by Jay on Dec 5, 2008 2:15 PM EST up reply actions

Thanks for the comments and kind words, everyone, I really appreciate it.

From this study, it does appear that Choo was rather lucky on balls in play, but he seems to have enough secondary skillz to still be quite valuable even if his batting average is .280 instead of .309.

by Peter Bendix on Dec 5, 2008 10:35 AM EST reply actions