- Tuesday, last of the eighth, game tied 2-2, two outs, runner on second, hit a triple to drive in the winning run in a 3-2 Pirates victory.
- Wednesday, last of the eighth, game tied 2-2, two outs, runner on first, hit a single to deep right to drive in the go-ahead run in a 5-2 Pirates victory.
Back-to-back games, he broke up tie games in the eighth inning with two outs. Clutch, right?
But how do we balance this against his overall performance? Polanco's one of the more disappointing Pirates hitters. Called up in June last year, he famously hit in his first eleven straight games, batting .365 with a .421 on base percentage and .442 slugging percentage. But then pitchers figured him out, and he batted just .204/.282/.320 the rest of the way. This year, the heroics against the Padres notwithstanding, he's hit .233/.306/.338. Among 77 players who've qualified for the National League batting title, he's 71st in batting average, 59th in on base percentage, and 70th in slugging percentage. So has he been a weak hitter who rises to the occasion with the game on the line?
The answer to this question has long been the province of speculation. Yes, there's evidence that, say, David Ortiz is clutch, given his postseason heroics. (Though did you know that in late and close situations--seventh inning or later with his team leading by a run, tied, or has the tying run on base, at bat, or on deck--his career OPS of .870 is 50 points lower than his overall OPS of .920? That spread is larger than average. So, relatively speaking, he's kind of choked late and close. I didn't know that, either, until I just checked.) But often, who's clutch and who isn't is a matter of reputation. The best example is the two best shortstops of the last 20 years, guys named Jeter and Rodriguez, and their respective polar opposite clutch reputations. Is there a way of measuring this, as opposed to just going by narrative?
Yes, there is. Here's how: When Polanco stepped to the plate Tuesday night, with Steve Lombardozzi at second and two outs, the Pirates had a 57% chance of winning the game. The reason I say that is this: Historically, when the home team is tied with two outs in the eighth and a man on second, they wind up winning 57% of the time. That's not some the output of some advanced mathematical model; it's based on actual games. The probability of winning, when you're the home team and there's two outs in a tie game and a runner on second in the eighth inning, is 57%. After Polanco's triple, the Pirates had two outs in the eighth, a one-run lead, and a batter on third. The probability of winning in that situation is 87%. So Polanco increased the probability of winning by 30%. Wednesday, the situation was mostly similar: two outs, tie game, eighth inning, but there was a runner on first instead of second. The odds still said the Pirates would win, but since the baserunner was on first instead of second, the probability of winning was a little lower, 55% rather than Tuesday's 57%. Polanco drove in the run and wound up at second after the throw home. So now the Pirates had a one-run lead with two outs in the eighth, but a runner on second rather than at a third, as they had the night before. The probability of winning, though, was the same, 87%, as it was after his triple Tuesday. So on Wednesday, Polanco increased the Pirates' chance of winning by 32%.
You can do this with every play of the game. Take Wednesday's game. Polanco led off the home half of the first, Pirates trailing 1-0. He grounded out, reducing the team's odds of winning from 43% to 41%. In the bottom of third, with two outs, nobody on, and the Pirates down 2-1, he flew out, reducing the chance of winning from 37% to 36%. He came up in the same situation in sixth and lined out to right, reducing the Pirates' chances from 31% to 29%. (Why did the odds go down from 37% to 31% in the same situation, i.e. two outs and nobody on? Because the Pirates had more of a chance of coming back in the third inning than in the sixth.) Then he added 32% in the eighth, ending the game plus 27%; plus 0.275, to be exact. That figure--the 0.275 Polanco added to the Pirates' odds of winning the game--is called win probability added, or WPA for short.
I think WPA is a fun number because it's a way of measuring clutch performance. Sticking with Wednesday's game, Polanco's go-ahead single added the most WPA, at 32%. But the second-biggest play was Neil Walker's double to lead off the Pirates' half of the seventh. At the time, the Pirates trailed 2-1. By getting into scoring position with no outs, Walker increased the Pirates' chance of winning, which was 34% with nobody out and nobody on, to 48% with nobody out and a runner in scoring position. He improved the Pirates winning probability by 14%. He eventually scored on a Francisco Cervelli groundout, but had he been stranded, he'd still get credit for increasing the Pirates' odds. That's what I like about WPA: It looks at every situation in a game, not just the ones on which runs score. If a player gets on base to put his team in position to win, he's helping the cause as much as his teammate who drives him in.
The clutchest Pirates game this year? It was the day before Polanco's heroics. In the second inning of Monday's Padres game, Pedro Alvarez grounded out with two outs and nobody on, the Pirates trailing 1-0. That reduced their winning probability from 39% to 38%. In the bottom of the fifth, still trailing 1-0, he grounded out again, this time with one out and nobody on. That reduced the odds from 36% to 34%. In the bottom of the seventh, still down 1-0, he came to the plate with two outs and runners on first and second. The Pirates' chance of winning was 33%. He got on by error, with McCutchen scoring and Cervelli going to third. Two outs, tie game, runners on first and third, home half of the seventh inning, the odds of winning jumped to 59%. So he added 26%. Then, in the last of the ninth, he came to the plate with two outs in a tie game, runners on first and third. The Pirates' chance of winning were 63%. Jung Ho Kang took second on defensive indifference, but that didn't materially increase the Pirates' odds; the lead runner, McCutchen on third, was all that mattered. Alvarez singled, winning the game, and increasing the Pirates' odds from 63% to, of course, 100%. Over the course of the game, he increased the Pirates' chance of winning by 63.1% and reduced it by 3.4% for a net of 59.7%, the highest Pirates WPA in one game this year. The worst WPA game of the year was Neil Walker's against the Twins on May 20. He grounded into a fielder's choice with none on in the first, grounded out with one out and nobody on in the third, lined into a double play with nobody out and a runner on first in the sixth, flew out leading off the eighth, struck out leading off the tenth, and lined out leading off the 13th. His 0-for-6 reduced the winning probability by 35.7% in aggregate.
Now, WPA isn't perfect. Players on bad teams don't get much of a chance to increase their win probability added, as they don't come to bat in close games. As the example of Alvarez shows, WPA gives as much credit for a single as an error--it's all about how the play ended, not how it got there. But it's better than nothing, and pretty easy to understand.
So how clutch is Polanco? No very, it turns out. His WPA for the year is 0.29--barely positive, 50th in the league this year. The team leader, unsurprisingly, is Andrew McCutchen, at 2.72. Alvarez is second at 1.35. The trailer is shortstop Jordy Mercer, at -1.38. The top ten in the National League:
Anthony Rizzo, Chicago 4.67
Paul Goldschmidt, Arizona 4.53
Bryce Harper, Washington 3.93
Kris Bryant, Chicago 3.15
Joey Votto, Cincinnati 2.73
Andrew McCutchen, Pittsburgh 2.72
Matt Carpenter, St. Louis 2.63
Todd Frazier, Cincinnati 2.26
Adrian Gonzalez, Los Angeles 2.22
Justin Upton, San Diego 2.18
In a way, this is a little unsatisfying, in that WPA is pretty much telling us simply that the best hitters are the best clutch hitters too. The order's maybe a little surprising--the two first baseman at the top of the list have a pretty big lead, and the rookie Bryant's been really good--but overall, you're looking at ten All-Stars or near All-Stars there.
Oh, and the Jeter vs. Rodriguez thing: The story is that Derek Jeter did little things to help the Yankees win: Get on base and move runners along, especially with the game on the line, while Alex Rodriguez did a lot of his damage in games that were already out of reach. WPA should capture that. A groundout to move a runner from first to second in the ninth inning of a tie game is more valuable than a three-run homer in a 8-1 game. What does WPA say?
Per 600 plate appearances, Jeter had 1.47 WPA over his career.
Per 600 plate appearances, Rodriguez has had 3.13 WPA over his career. That's more than twice Jeter's rate.