The Baseball Graphs Blog

Friday, October 20, 2006


When I saw that catch by Endy Chavez, I figured it was fate.  I mean, I know I’m supposed to be a baseball analyst and all, but I’m a fan first and foremost.  Well, from a WPA viewpoint that catch was the second-biggest play of the game, behind Molina’s home run.  But back to the fan thing: what a great start by Oliver Perez.  With Perez in a groove and Chavez channeling Tommie Agee, I really believed the Mets would win, right up to the Molina homer.  Then I changed.

Even when the Mets loaded the bases in the ninth, I wasn’t there.  My analytic mind had taken over, knowing that their WPA with the bases loaded, two out and two runs behind was only 17%.  Yes, the catch made me a fan, but the home run turned me back into a geek.

It’s ironic, I think, that the Mets’ flaw in this series was their offense, not really their inury-racked starters.  I’m personally hoping they move Heilman to the rotation next year, a role in which one floating changeup might not hurt as much.

Ah, I’m rambling incoherently.  I’m in shock.  I’m depressed.  It’s 5:00 a.m. and I can’t sleep.  For a more literate and coherent Mets’ requiem, I recommend Jessica’s blog entry.  Maybe I’ll go back to bed, or maybe I’ll just get back to work on the THT Annual.  Either way, I won’t feel quite the same as I did a few hours ago.  There will be a little bit less to look forward to.

Following Jessica’s lead, let me be the second to congratulate the Mets’ on their 2007 World Series victory.

Posted by Studes on 10/20 at 03:52 AM
Tuesday, October 03, 2006

WPA Curiosities

I logged today’s A’s-Twins game in my WPA spreadsheet and found some curious things.  Frank Thomas was voted the Game MVP (rightfully, in my opinion), but he actually finished third on the A’s in WPA, behind Zito and Street.  Thomas’s home runs occurred in less critical situations: the second inning, when the LI was just 0.89, and in the ninth with the A’s up 2-1 (LI of 0.70), so the WPA of each home run was just under .10 in both cases.

There were several bigger hits in the game, including Scutaro’s double in the second (.103) and Jason Bartlett’s double in the eighth with none out and the Twins down by a run (.166).  Also, the flyball that Milton Bradley lost in the Metrodome roof was worth .166 WPA, too.  Those were the two biggest plays of the game.

So Frank Thomas, who homered for two of the A’s three runs, didn’t lead the team in WPA and didn’t even have the biggest hits of the game.  Go figure.

Posted by Studes on 10/03 at 03:17 PM
Win Probability AddedPermalink
Monday, October 02, 2006

WPA Usage

Nate Silver has introduced a new stat at Baseball Prospectus called Secret Sauce.  It combines those elements Nate and Dayn Perry found have the most impact on postseason success: fielding, strikeouts and bullpen.  It’s an interesting stat, but I found it disconcerting that Nate used WXRL (closer only) for his bullpen measurement.

WXRL is Baseball Prospectus’s version of WPA.  As always, care should be taken when interpreting WPA and I don’t think Nate gets it quite right.  In this case, WXRL includes both the performance of the ace and the number of close situations he was brought into.  Closers have no impact on the criticality of their appearances—that is totally up to the manager.  I don’t understand why Nate would include it.

Now, Nate might say that WXRL came out well in his regression analysis, but that wouldn’t be a good response.  Even though a variable may work well in a regression analysis, that’s not a good reason to include it in a formula.  There needs to be a logical rationale for why it should be included.  I don’t see the logic behind using WXRL.

On another note, the Wall Street Journal ran an article last Friday (by Sam Walker, who mentioned our site in his book Fantasyland) that ranked the clutchiness of all postseason plays.  Guess what they found to be the biggest clutch hit?  Tony Womacks’ game-tying double against Mariano Rivera in the bottom of the ninth, Game Seven, 2001.  The one before Luis Gonzalez’s game-winning hit.  Yes, that was deemed even clutchier than Bill Mazeroski’s home run against the Yankees in 1960.

It’s an interesting read, and Walker is to commended for using WPA as the baseline for his analysis.  Unfortunately, he included several other elements in his analysis.  Actually, the elements (quality of the pitcher, impact on probability of winning the series and not just the game) were good, but the approach was kludgy.

Oh well.  Still an interesting read.  Here’s the link, though I think you need to be a subscriber to read it.  Hat tip, JP.

Posted by Studes on 10/02 at 01:23 PM
Win Probability AddedPermalink