THE BOOK--Playing The Percentages In Baseball

I sent a long e-mail to Dave Studeman last night about this (they posted it on THT Live). I don’t get why people are looking at quality of team instead of individual batters faced. Baseball Prospectus calculates it all for us on an individual batter level (click name).

If you use the 1.73OBP+SLG method that Tango outlined a little while back, the difference between Lee’s opposing batters and the average AL (including interleague) batters was something like 5 runs.

207 expected runs for batters that Lee faced

(1.73*.327+.404)*.27*792=207.4

And 212 expected runs if the batters that Lee faced were league average.

(1.73*.333+.418)*.27*792=212.5

The same for Halladay (scaled to Lee’s TBF of 792) gives you 217.2 runs.

So the difference for each is about 5 runs from the mean. I don’t think that’s the reason for Lee’s outstanding season.

While that is not the reason for Lee’s outstanding season, that is a huge difference!  Almost half a run in runs allowed per 9, around .4 in ERA.

The 1.73*OBP+SLG times all that stuff does not equal to runs, but is related to runs at a 1:1 level.

Specifically, this:
(1.73*.327+.404)*.27*792

should be:
(1.75*.327+.404-1)*.27*792

And it will give you runs above/below average.  So, in this case, it’s -5 runs.

It’s the same result, but I don’t want anyone to get confused about the gross numbers being reported.

Anyway, it’s a difference of 10 runs between the two, which is fairly substantial.  However, using current performance stats is not the best way to get the true talent levels of the opponents faced, but it would be fairly close.

I have to profess ignorance. What exactly does the (1.73*OBP + SLG) * 0.27 * PA tell us? The below average batters that Lee faced would’ve scored over a run an inning (207.4 runs in 201.7 innings) against an average pitcher?

The other day I took the average runs scored by their opponents reported by Batter’s Box and figured Lee was 49 runs better than average, Halladay about 41 or 42.

It closes the gap, but Lee still has the advantage.  But if we go there, shouldn’t we look at defensive support as well?  Baseball ref has Cleveland at .690, and Toronto at .703 in defensive efficiency.  THT has Toronto at .706, Cleveland at .691, not sure why the numbers don’t match.  THT also shows Toronto’s defense at +66, best in the league, with Cleveland at +20.  (I think those are plays saved, not runs)

Lee is responsible for about 14.5% of Cleveland’s batters faced, Doc 16% for Toronto.  Looks like Scott Rolen, John McDonald & co. have been worth an extra 5 runs to Halladay, making up for the tougher batters he’s faced.

According to Situational Wins (WPA/LI), Cliff Lee is +5.1 wins and Halladay is +5.0 wins.

Does Situational Wins work as well for pitchers as hitters?  No.  The reason is that WPA/LI treats each PA as identically weighted, that each have the same impact.  However, pitchers control their base/out situations, while hitters find themselves in various game state situations.  It’s fair to put each batter on the same scale, but not pitchers.

I’m not sure how to best resolve this issue.

***

The best relievers by the way in Situational Wins are Kuo in LA, Scott Downs with the Jays, the ageless Mariano Rivera, and Soria.  K-Rod is way way down, clearly the effect of opportunities over performance.

More from Sheehan:

http://www.baseballprospectus.com/chat/chat.php?chatId=519

BL (Bozeman, MT): Hi Joe, thanks for the chat. Given the level of variance and randomness you detailed in your Prospectus Today look at the Dodgers, how useful is a comparison like yours in Unfiltered Halladay and Lee?

Joe Sheehan: I assume the thought is that teams vary in their performance enough that evaluating opposition by overall performance isn’t valuable. It’s possible that Cliff Lee faced teams when they were playing well or something. It’s not an unreasonable position, but I think you have to take the position that seasonal stats are the best information we have for judging teams, and therefore, adding granularity increases complexity without necessarily adding information.

I did get asked a few times about the Opponents’ Batting stat report, which shows about 35 points of OPS edge for Halladay’s opponents. This is additional information, and useful, but I shy away from using that report a bit as the range of figures there is fairly small. And as I say, I don’t know how much weight you should give this information in evaluating the two pitchers’ seasons.