WPA and the Play of the Game
by jscape2000 on Jun 10, 2010 1:00 PM EDT
in New York Yankees Analysis
Email
PrintLast night, Robinson Cano made a play on a hard hit grounder off the bat of Nick Markakis, diving toward first base on the outfield glass to keep the ball from rolling into right field. Cesar Izturis took off from second base representing the tying run, but he was forced to stop at third.
According to FanGraphs' Win Probability Added (WPA), that play increased the Orioles' chances of winning by 3.5%.
From an AB to AB approach this is true: the Orioles were more likely to win with runners on 1st and 3rd than with a runner on 2nd. But for all the crisp clarity of play-to-play, there are moments within the play that break down farther than WPA can calculate.
After getting ahead in the count 1-0, Markakis watched a curveball strike then fouled off a fastball. In a 1-2 count, he took an emergency hack at a curveball that Gameday shows as low. Sabathia has recorded swinging strikes on only 8.3% of his pitches, well below his 10.7% career average, and Markakis is one of the better contact hitters in the league. True to form, Markakis hit it well.
There was never a play on the ball for Cano. I've seen a few 2B who I think could have turned that grounder into an out: Orlando Hudson when he was with Toronto, Roberto Alomar in his prime, maybe a few others). But it would have been a great play for any second baseman, and I was prepared to watch that ball roll into right field.
When Cano dove, he could have missed it. The ball could have deflected off his glove and rolled away. When he popped up and threw to first, he could have airmailed it or Tex could have missed the throw. Sabathia could have given up a 3 run homer to the next batter and made the great play a moot point.
If the runner scores, the play increases the Orioles chances of winning at that point in the game by 22.2%. That's a bigger swing than the possibility of Cano's go ahead RBI single in the 6th (13.3%), bigger than Sabathia's subsequent K of Luke Scott with the bases loaded to get out of the jam (15.4%).
WPA is light years beyond the traditional boxscore- but there are some plays that are great plays that just don't show up in the spreadsheet.
9 comments
|
0 recs |
Do you like this story?
Comments
It was the play of the game.
The play increased the Orioles’ chances of winning by 3.5%. If the runner scores, the play increases the Orioles chances of winning at that point in the game by 22.2%. 22.2% – 3.4% = 18.8% savings.
by PinstripedHippie on Jun 10, 2010 1:12 PM EDT reply actions
Great post jscape
I agree, we can always find better metrics too, but what PinstripedHippie says makes perfect sense, and was the first thing that came to my mind too. Preventing a rise in opponent WPA is technically the same as adding to the probability of a win.
by pkyankeefan on Jun 10, 2010 1:23 PM EDT via mobile reply actions
Thanks
There must be a way to break down WPA to represent the potential of balls in play, maybe by mixing it with the batted ball LD/GB/FB data.
I think it’s the next step for sabermetrics.
"Have faith in the Yankees, my son. Think of the great DiMaggio."
There's "PZR".
Which is basically the first half of UZR. It’s “how likely is this batted ball to be turned into an out”. That gets credited to (or blamed on) the pitcher. If a fielder makes the play, he gets the difference between the PZR and making the play. If he doesn’t he gets the difference between PZR and not making the play. For example, if a fly ball to the OF is caught 90% of the time, the pitcher gets credited with .9 outs and .1 hits (you can also stretch that to what the game would look like if an out or hit happens). If the outfielder makes the play, he gets credit .1 outs. If he misses it, he’s penalized .9 outs.
So, after that long story, if you know the probability of every batted ball being an out or hit, you can dole out partial credits to pitchers and fielders.
just goes to show
stats are useful, but they need to be taken in context. no stat ever tells the whole story.
עם ישראל חי
technically speaking, like you say the result of the play still increased the orioles odds to win, since no out was recorded and the runners moved up – ofcourse it was a great play – but based on this metric, the play wouldn’t make their chance to win go down even though he made a great play so i dont know really what the point of the argument is – if i was an o’s fan there i’d be disappointed but atlest you still get another chance with the next hitter and the 3.5% represents that
I think you've stated the point of the argument.
“[B]ased on this metric, the play wouldn’t make their chance to win go down even though he made a great play….” In other words, WPA doesn’t always indicate which plays are great, or even the most important plays of the game.
I like the idea of using potential outcomes avoided to quantify how great a play is, perhaps something like (p-1)PWPA, where p is the probability of making the play and PWPA is the WPA if the play is not made. (Of course, PWPA itself would have to be based on how likely each of the possible outcomes are if the ball is not played.) This stat would require a definition for “making a play” that doesn’t necessarily involve recording an out.
Measuring the importance of such a play seems a little more difficult. As jscape pointed out, that all depends on subsequent events. The same can be said about WPA, though, as tonight’s play of the game demonstrated. It’s worse when you get into potential outcomes: how far do you string out the potential damage that was avoided?
Great, thought-provoking post, jscape.
Usqueadbaugham! Anam muck an dhoul ! Did ye drink me doornail?
Righty vs lefty
Maybe someone can explain this to me ( I’m serious) Last weekend in Toronto The Yankees faced Lefty Ricky Romero During the game the announcers make a comment several times that lefties are hitting over .300 against him.
This week in Baltimore, The Orioles brought in left handed relievers, against whom Left handed hitters are hitting almost .400 yet ever single time the Yankees switch hitters batted Right. Girardi and most of MLB is consumed by numbers and stats. I don’t get it
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Community Guidelines
Related FanPosts on Pinstripe Alley
Related FanShots on Pinstripe Alley
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
