Win Probability: The suspense factor and other errata

The College World Series began last week with the usual set of great games and inspiring teams, but with a little wrinkle to the ESPN broadcast that should probably interest MLB fans: the inclusion of win probability to the bevy of stats used over the course of the broadcast.  I’ve already blogged about it twice (Part 1, Part 2, and once since: Part 4), but it’s a discussion that warrants continuing because of the widespread influence of the Worldwide Leader on casual baseball fans.  If they’re doing it right, ESPN could bring the discussion of win probability to the forefront of sabermetrics.  It certainly wouldn’t hurt if people like Tangotiger, Keith Woolner, Studes, and David Appelman, get a little increased visibility and/or credit for their work in this area.

I’ve been tracking some of the reactions to win probability around the blogosphere, and I thought that warranted a separate post.  Here’s what I found:

First, a link to a short exchange I had with Tango on the FanGraphs forum.

Just Call Me Juice has an interesting perspective about win probability:

The second thing is something not completely related to baseball, but to the brains behind the ESPN broadcasts. If you have been watching (and perhaps this has been done for other games and sports, but I haven’t noticed it), you may have seen the “Win Probability” statistic flash across the screen a few dozen times per inning. This is basically a percentage that shifts throughout the game based on situations (inning, score, outs, runners on base, etc.) that tells the viewer who is “supposed” to win the game at any given point. Interesting concept? Yes. Compatible with the sports world? No. This is a terrible idea and I really hope that ESPN discontinues it. Sports are all about improbabilities. This is what keeps us viewers intrigued and watching on the edge of our seats.

The author elaborated further, and although I don’t think it warrants excluding the stat or saying that proponents of win probability are “degrading” anyone’s performance or “not true sports fans” (as he later says), he makes a good point about the “suspense factor” of the broadcast.  In the game I reviewed, Louisville blew out Mississippi State, and ESPN wisely only showed the graphic twice.  However, I don’t think that knowing the probability actually makes improbable feats any less impressive.  It may even work the other way, which is probably the only reason ESPN considered it in the first place.

Here’s another potential problem, as elaborated by a fairly well-informed user at the college baseball message board (user: ScoreInjected):

The statistic in and of itself is not right or wrong and may not be bogus if it isn’t unreliable. The probability is what it is. What is bogus is if ESPN has not explained the algorithm (or even if they simply have not made that explanation reasonably easy to obtain). That’s stupid, because “what it is” is something we as as fans don’t know. Since we don’t know the algorithm and we don’t have a history of its reliability, it is worthless to us. Either by itself would give it some value to us, but we should have the algorithm now. What is stupid is trying to use any statistic without knowing how it is derived because you don’t know what it is.

Like most of the viewers, this person wasn’t aware that there are people tracking win probability for baseball and that it’s not a new concept.  That’s nothing new, I suppose.  What I don’t want to see is ESPN over-promoting this idea during the CWS and eventually bringing it into the MLB broadcast without actually giving it a more detailed on-air explanation.  There are some purported criteria, but that’s all they’ve given thus far during the broadcasts.  Based on my analysis, they’re obviously not just making these numbers up, but there’s a good bit of variability between different methods of tracking win probability, and they ought to come out and explain their algorithm and/or tables.  Nerds like me want to know.

Awful Announcing has predictably slammed ESPN for the win probability feature without really giving it a good analysis.  Their schtick is funny most of the time, but this mostly just came across as uninformed to me.

Mariner Optimist beat me to the punch with their announcement of win probability as an upcoming broadcast addition last week.  They see it as a future broadcast staple, so that’s kind of a different perspective from everyone else around the web.

I saw a handful of other perspectives out there, but that covers pretty much everything for now.  I’ll stay tuned and see if they go into it a little deeper on ESPN.


2 thoughts on “Win Probability: The suspense factor and other errata

  1. Another problem with ESPN’s decision to show WP is this: When you’re trying to introduce a new idea to people who don’t necessarily want to hear about it, one of the worst things you can do is to introduce it in such a way that it’s easily dismissed.

    If ESPN just shows the number, without going into both its origins and limitations, then that’s exactly what happens. People won’t think it’s worthwhile. And what’s bad is that then, later on, when somebody tries to better explain the concept to them, they’ll think, “I’ve already heard about WP and it’s stupid. I don’t want to hear about it anymore.”

    It’s similar to the tragic tale of computer rankings in college football. When they first got popular in the 90’s, many people dismissed the whole idea as stupid, partially because the systems were almost never explained well. (And also partially because many of the systems ARE stupid and produced bad results, but that’s a different story.)

  2. About 80% of the things I’ve read about ESPN’s introduction of win probability have been negative, with most of them calling it a gimmick, an attempt to borrow from the WSOP, or something along those lines. Probably over half of them didn’t make a better point than “people don’t want this,” which at this point is a totally unsubstantiated claim. I won’t say the opposite is true, but the idea at least warrants some real consideration.

    I hadn’t thought about the angle that ESPN’s poor introduction of win probability-based stats might be the nail in the coffin for WPA before it even gets off the ground. They need to turn this around by introducing it well to an MLB audience. Do a Baseball Tonight segment on it. Anything would be great, as long as they don’t let Joe Morgan handle it.

    I’ve been doing my best to keep following up on this because it’s driving me up in the Google search rankings for “ESPN win probability” and other related keywords. My intent is to get a reasonable voice out there as one of the top results for people who are trying to figure out what ESPN is doing. So far, it’s working, but I can only use the same argument so many times before the popular sites start passing me up.

    Now, a Morgan-related aside…during last night’s Tigers-Braves game, Joe defended Gary Sheffield’s remarks about Latin players by saying he read Sheffield’s book to understand the context of what he was saying. The argument mostly made sense to me, but it’s almost comical because he still won’t bother to read Moneyball to get the big picture about sabermetrics. He’s leaving no doubt that his personal biases have completely clouded his thinking beyond repair.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s