THE BOOK--Playing The Percentages In Baseball

My only complaint is no one corrects Harold Reynolds’ contention:

“But it penalizes pitchers for pitching around guys!”

Yes, it does.  But if the pitcher then gets the next guy out, the guy he wanted to face (instead of the guy he pitched around), it EXACTLY COMPENSATES for the loss of WPA of the pitched-around player. 

Sigh.

They also didn’t discuss the difference between a predictive stat (tRA, FIP, etc.) and an evaluative stat like WPA.  To paraphrase, one of the participants states that MLB teams aren’t looking at WPA when figuring out which relievers to go get and attempts to use this fact to somewhat discredit WPA.  Of course they aren’t acquiring players based on WPA.  WPA includes all manner of luck.  It isn’t designed to predict future performance, it is designed to evaluate past performance.  I can’t think of a better way to evaluate a team’s outcomes while a certain pitcher was in the game.  Yeah, a high WPA could be largely driven by luck, but it happened nonetheless.

The thing is that Ken was all over saying why he didn’t like it too, but he didn’t even make any sense.
Hey, let’s just use batting avg. with runners on to settle everything ???

I would say that given a 5-minute conversation, and the way they were speaking that they could easily buy into the concept more.

They showed great willingness, and they just need a bit more of an education.

Really, what they said, and how they said it, I couldn’t ask for more.

I agree that it was a pretty good discussion given the personnel, context, etc.

I was a little surprised at the level that Ken and Tom understood the concept.  In all fairness to them, it is a little difficult to articulate the strengths and weaknesses of WPA in that venue. I think that understanding and explaining the differences between the explanatory and predictive aspects of a metric is beyond the capabilities of Rosenthal and Verducci.

Do we need any more evidence that Reynolds is clueless, at least on the statistical analysis front, and that his status among baseball “analysts” is grossly overrated and overstated?  He was at least willing to listen and I don’t hold his lack of knowledge in that regard against him as long as he does not attempt to debate someone who is discussing something to do with statistical analysis or sabermetrics in general.

Hm… I didn’t like it at all.  I didn’t like that they emphasized it as a “reliever valuation tool”, I didn’t like that they didn’t mention anything about it not being very predictive, and I didn’t like that they kept bringing up the stuff about pitching around a batter.

Three guys having a fairly intelligent discussion, with Harold Reynolds still searching for a clue.

It could have been better if they had a guest who was really knew the subject and could defend it against their questions.

Yeah, it’s kind of a pointless endeavor if you have four guys talking about something they know very little about and trying to understand it on the spot.

I think Harold would’ve bought the stat more if they also talked about its use for hitters. (Unless he says it doesn’t take into account “clogging up the base paths.")

While I really appreciate MLB Network’s efforts to incorporate and discuss sabermetric stats (and while this specific discussion was fairly acceptable), it’s not like they don’t have the means to have either the creator of the statistic or a person knowledgeable enough in the subject matter to present the concept and answer each of the other analysts questions.  They’re already dedicating 5 minutes of air time and 4 analysts to it.  When is comes to sabermetrics and the mainstream media, I’m of the opinion that they should either do it right or leave it alone.  I assume they’re incorporating it because of a perceived demand for it.  I know I’d watch more MLB network if Tango or Cameron were making an appearance!

This is from a few days ago, but I was watching Olympic curling the other day, and the broadcasters brought up win probability.  It was something like one team was up 1 with the hammer in the 9th end, and the end was shaping up to lead to a choice of whether to play for a blank and keep the hammer in the 10th, or give up the hammer and take a 2 point lead into the 10th.  One of the broadcasters asked his partner (who was apparently a competitive curler) whether teams generally preferred to play to keep the hammer protecting a 1 point lead or to take the point in that situation, and she responded accordingly.  Then, they pulled up the numbers and said that teams win x percent of the time down 1 without the hammer in the 10th and win y percent of the time down 2 with the hammer in the 10th (it was something like 10.4% for both scenarios, I think).  I don’t know if their data is just from Olympic matches or if it includes only the top bonspiels or where the cutoff was for data to include.  Anyway, it was pretty cool to hear win probability used to describe decision-making in such a strategic sport as curling (which is my favourite use of win probability).