THE BOOK--Playing The Percentages In Baseball

Wednesday, November 12, 2008

Will the real Chipper Jones please stand up?

By Tangotiger, 02:29 PM

Back when Chipper was hitting .420, Nate Silver gave him a 13% chance of hitting .400. He based that on Chipper being a true .348 hitter. I used Bayes to come up with him being a true .321 hitter. Marcel, at the start of the season, forecasted him to be a true .307 hitter.

From the point when we all made our bets…
.348 Nate Silver
.330 Marcel (includes using 2008-to-date of Jun 11 performance)
.321 Tango (via Bayes, using only 2008-to-date of Jun 11 performance)
.307 Marcel (excluding any 2008 performance)
...this is what Chipper Jones did: he went 68 for 220, for a .309 batting average. Seeing that a .3072 batting average on 220 AB gives you 67.6 hits, which we get to round to 68 hits, the dumb monkey nailed Chipper’s batting average perfectly.

Chalk up another win…

(Not that it really means anything. After all, as we bore witness, anything can happen in 219 AB, and so, anything could have happened in 220 AB. 92 hits here, or 68 hits there, all from the same guy just months apart, is not a big deal.)

• Sabermetrics • Forecasting

#1 (see all posts) 2008/11/12 (Wed) @ 20:33

Well, Chipper *was* playing hurt for muh of the remainder of the season (when he played). I don’t think it’s quite a fair comparison.

#2 (see all posts) 2008/11/12 (Wed) @ 21:16

But how can you predict a player’s 2nd half performance by ignoring what he did in the 1st half? It should at least be part of the mix.

After 2007, I have him projected at .310. Now, .323 (without age adjustment)

#3 Sean (see all posts) 2008/11/12 (Wed) @ 21:24

Agree with Brian. I don’t think you can chalk up a win based the .307 Marcel, using that version you were ignoring data. Now the .330 Marcel and .321 Tango were still closer than Nate’s .348, but not as close.

#4 tangotiger (see all posts) 2008/11/12 (Wed) @ 22:27

I noted as much on Primer, which I’ll repeat for those who missed it:
http://www.baseballthinkfactory.org/files/newsstand/discussion/the_book_blog_tangotiger_will_the_real_chipper_jones_please_stand_up/

Yes, of course, if you were to bet, either the “smart” Marcel pick (.330) or the Bayes pick (.321) would have been the one to go on.

The best pick, as it turns out, was the “dumb” monkey pick of .307, that completely ignored the .420 batting average through Jun 10.

And like I said, it doesn’t really mean anything…

#5 tangotiger (see all posts) 2008/11/12 (Wed) @ 22:32

As for Dan’s contention about him being hurt: that’s part of the forecast. You don’t make a forecast presuming 100% healthy. The forecast accepts a certain probability that he’s going to play hurt, and whatnot. It’s not like Nate said that Chipper has a 13% chance of getting to .400, if he’s not hurt.

The only thing that’s not “fair” is that I cherry-picked this one guy. Which is why I’m in the middle of working on analyzing the picks of a bunch of forecasters for 2008 (Marcel, Bill James, MGL, Pecota, Zips, Chone, THT, Sackmann). We’ll see how that turns out.

#6 David Pinto (see all posts) 2008/11/14 (Fri) @ 11:35

Earlier in the year I was running charts on Jones’ probability, using .310 (his career average at the time I started the chart. Lots of people criticized me for using a low number, so I’m glad to see some vindication.

Page 1 of 1 pages

<< Back to main

THE BOOK--Playing The Percentages In Baseball

Wednesday, November 12, 2008

Will the real Chipper Jones please stand up?

Latest...