THE BOOK--Playing The Percentages In Baseball

I actually had forgotten I did this interview, and checking my inbox, I see it was done less than three months ago.  It is bothersome how I can remember stuff I read from Bill James from 25 years ago, and I can’t remember my own stuff from just months ago.

Anyway, for those interested, here was everything I said, and you’ll have to imagine what Peter asked me:

***

I’m not sure we’ve seen [breathless claims] these past few years, likely because of Fangraphs leading the charge. Baseball-Reference has also embraced it. Before them, yes, it did seem some people thought this was brand new.

***

[Mills brothers] are the first that I know of. I doubt it would go back much further, if only because of the data required (machine intensive).

***

Fangraphs and Baseball-Reference both use win expectancy numbers that I have provided. So, I like both. Is there anyone else doing this? If so, I haven’t seen it to comment on.

***

Right, that’s exactly the description we need. It’s a story stat, as Studes at HardballTimes calls it. It captures the qualitative aspects of how you feel at the moment the event occurs, and quantifies it. Once you do that, you can start to compare across games, or seasons. It makes looking for wild games much easier.

***

Yes, I think so [that we need a good presentation]. The day after the Bartman/Cubs collapse game, I introduced my first WPA story game: http://tangotiger.net/archives/stud0163.shtml

That game was as tense as any game I’ve ever seen. How can you capture that game with numbers? And I think only WPA could do it. A few years later, I introduced it in graphical mode: http://www.hardballtimes.com/main/article/crucial-situations

And I think, that’s how it’s sold. Those who are open to the idea, love it. Those who aren’t, just won’t care.

***

I wouldn’t do it for every pitch. I’d probably only bring it up when the starting pitcher is in trouble in the 5th or later innings, and when relievers enter the game. You have to pick your spots. It works with poker because you go from 50% to 0%/100% within 2 minutes. Seeing a graph chart after every pitch going from 50% to 0/100 within 3 hours will lose its impact.

***

WPA should be used to try to capture the clutch aspect of play. For example, it’s WPA that can show you the difference between the clutch play of Ortiz and the non-clutch play of ARod when they both had, on the surface, two great years.

***

I just read this, so, yeah, I agree with you [to be used in MVP].

***

Sure, absolutely [for hockey and soccer]. I have no doubt that sportsbooks where betting is legal in Europe definitely use win charts for soccer. They’d have to. I have a win chart for hockey, which I have not published.

Thanks Tango!I think WPA is growing rather rapidly compared to other stats, and Peter lists some solid reasons for the appeal. While most mainstream “advanced stats” are stuck on VORP and other now useless measurements, WPA has started to emerge in that group. It was also featured in MLB 2K11 (sponsored by Pepsi!), and I think education through video games is a very effective way to communicate the usefulness of more complicated statistics to the mainstream.

I think Fangraphs is probably a lead reason for its popularity.

I love the poker analogy.  Those percentages during the hands are about as key to enjoying those broadcasts as the yellow line is to football.  Not required, but a BIG step in the right direction.  I’d love it if it popped up during baseball games.  Or any game, for that matter.

I think I’ve suggested it before, but I think Fangraphs could make money marketing t-shirts of some of the more interesting graphical presentations of WPA for significant games.  What Mariners fan wouldn’t want a graph of Game 5 of the ALDS?  What Blue Jays fan wouldn’t want Game 6 of the 1993 World Series?  We can all fill in our favorites, but that’s what’s so fun about WPA.  Any fan can get it instantly when it’s presented graphically.

One interesting thing about WPA is that it evaluates relievers very differently from WAR for example.

I know of many people who would use WPA for relievers instead of WAR, simply because it matches up better with their evaluation of closers and relief aces. It’s simpler to believe that Rafael Soriano added 4 Wins instead of 1.6 (WAR), for instance. It’s also easier to believe that Rivera has 51 WPA against 36.5 WAR in his career. 14.5 Wins is nothing to sneeze at.

But if you value relievers using WPA, doesn’t that reward them for just pitching in higher leverage situations (ie something determined by their manager not their ability)?

Jono, they still have to pitch well, or they would get a big minus WPA.  So yeah, WPA rates a scoreless 2nd inning in a tied game as less than a scoreless 9th in a tied game.  But that makes sense doesn’t it.  It is a combination of the context determined by their manager and teammates, (and opposition), and their ability to do well in the more important context.

Huh, what a coincidence, I already have a response written up. How I feel about using win expectancy to evaluate relievers:

http://www.baseballprospectus.com/article.php?articleid=12747

Coincidence back at you, as I posted your article, and my thoughts on the issue of relievers and leverage just a few minutes before your post!