Win Probability Added:What is it good for?
Posted by The Right Reverend on April 25, 2006
I'm sure most of you have noticed the daily "Win Probability Added" charts copied and pasted from the wonderful Fangraphs.com website. And I'm sure some of you have took a look at it, scratched your head, said "what the heck is this?" and moved on. So, let me do my best to explain what the charts mean, what Win Probability Added is, why it's good, how it's not perfect, and all that other stuff.
First of all, I'm not sabermetrician, mathmatician, statician, or any other sort of "ician". But I've enjoyed baseball stats for as long as I can remember, looking at the backs of baseball cards and pouring over the Sunday sports section stat columns. I'm not sure why the fascination, but I always enjoyed that sort of thing, and when I picked up the book Moneyball, it was like a whole new world opened up to me. Since then, I've been an avid reader of publications like Baseball Prospectus, Ron Shandler and The Hardball Times. But I admit, I don't know how they come up with a lot of their complicated formulas, though I've tried. Last night, I tried to read Keith Woolner's article in BP 2005, titled "An Analytical Frameword for Win Expectancy", and while I got some of his points, I felt like I was listening to The Architect and I was a dumbed-down version of Neo who couldn't save Zion, Trinity or even $10.
Anyway, on to my dumbed-down version of understanding Win Expectancy, or Win Probability (same thing.) Basically, it can be used to track the ups and downs of the game. It helps to assign value, positive or negative, that a player has on the game. Each team starts with a 50% chance of winning the game. Throughout the game there are different baseball "events." Ie, an RBI single, a sac-bunt to move the runner over, an infielder's error or a blown call by the home plate umpire. These events, depending on other circumstances like what other runner's are on base, what inning it is, etc, effect a teams percentage of winning that game.
Taking a look at 4/16, aka the "Albert Pujols Easter Sunday Extravaganza", I'll try an illustrate this. When the Cardinals came to the plate, they were down by one run and had a 48.7% chance of winning the game. With 2 outs, runners at first and second and John Rodriguez at the plate, the Cardinals then had a 48.2% chance of winning the game. Rodriguez promptly tripled, clearing the bases and giving the Cards a 2-1 lead. Well now, they had a 67.4% chance of winning that game.
Or another (painful) example, Encarnacion's dropped ball. It was the 7th inning, the Reds are down by 2 after Quinton McCracken's home-run. The Reds then had a 22% WPA. Ryan Freel walked, 31.4%. Felipe Lopez then hit a routine fly to Juan, putting runners on first and third. That play then gave the Reds a 61.8% WPA. Nice job, Juan (-31.4%) Rich Aurilia promptly doubled 85.8% WPA.
When Jason Marquis stepped to the plate, the Cardinals only had an 18.8% WPA. Marquis's single gave the Cards a 31.5% WPA and then Pujols homerun off course won the game, 7-6. So one swing of the bat gave Pujols a +68.5% contribution to the win.
Basically, what WPA does is tell the story of the game It explains in graphs and charts what's happening and it's value, and helps you understand some of the "what" that's going on in the game. It does have flaws. First, it has no predictive value. It simply looks at what's happened in the game. It also can be dicey on how much you want to attribute on a particular fielding play. There are other flaws, of course, but for now I'll just refer you to this article for now.
Bottom line, it's all about the narrative of the game. It tells the story of the game in an interesting way, though it may seem a bit nerdish to some. When the chart is up, the fans feel up. When the chart is down, the fans feel down. And when Pujols hits a monster shot to deep left field to win the game, the feelings go from being down to euphoria.