Introducing pWin: A Better Pitcher Decision Statistic (Part 2 of 3)

Introducing pWin: A Better Pitcher Decision Statistic (Part 2 of 3)

In my previous post, I argued that the Win statistic in baseball is too arbitrary and introduced a new statistic to better reflect a pitcher’s contribution to a team’s victory: the pWin. This stat is given to the pitcher who contributed the highest WPA for the winning team. Let’s look at this a bit closer using Johnny Cueto’s 2015 season. How would Cueto’s pWin/pLoss record compare to his traditional W/L record?

Using the traditional W/L method, Cueto was 11-13 in 32 starts. Using pWin/pLoss, he would have been 13-10. This is not a radical departure from his traditional record, but I think all who followed Cueto last year would agree it is more in line with what he produced. Let’s look at the details and also address an issue with assigning pLosses.

In 27 of the 32 Cueto starts, both the traditional W/L and pWin/pLoss systems have the same result, so we’ll just ignore those for this analysis. Let’s look at the other five:

Pirates vs. Reds, April 6, 2015 (Opening Day)

See part 1 of this series for an analysis of assigning Cueto the pWin for this game, even though he didn’t receive the W.

Cardinals vs. Reds, April 11, 2015 (Cardinals win 4-1)

Screen Shot 2016-04-08 at 10.58.23 AM

In this game, the Cardinals beat the Reds 4-1. Cueto got tagged with the Loss for giving up 2 runs (only 1 earned) in 7 innings. However, Burke Badenhop had the lower WPA: -0.086 vs. Cueto’s 0.107. So what happened?

Cueto left the game after 7 innings with the Reds down 2-1. However, Badenhop came in, downed the Cardinals 1-2-3 in the 8th, but then gave up two runs in the top of the 9th. Manny Parra then had to come in to clean up the mess.

Traditionally, this is clear-cut: Cueto gave up more runs than the Reds would score, so he should get the Loss. But looking at the overall game, Badenhop gave up just as many runs (and both of his were earned runs), so why should Cueto get the Loss for giving up 2 runs in 7 innings, while Badenhop doesn’t for giving up 2 runs in 1.1 innings? The timing of the runs doesn’t really matter to the final result. So it seems to me that the pLoss should go to Badenhop and Cueto should have a no-decision.

Giants vs. Reds, May 14, 2015 (Reds win 4-3)

Screen Shot 2016-04-08 at 11.05.02 AM

The Reds won this game 4-3 against the defending World Champions, with Jumbo Diaz (again) getting the Win for a short night of work (2/3rd of an inning). However, Cueto should get the pWin, as he amassed 0.170 WPA vs. 0.105 for Diaz (Chapman had 0.154).

This is a classic example of the weakness of the traditional Win statistic. Cueto pitched 7 innings, giving up only 2 runs in a game in which his team scored 4, and he even left with his team winning 3-2! Yet he gets a no-decision because Tony Cingrani came in and quickly gave up a run to tie it up. When the Reds came back, Cueto was out of the game and so Diaz got the W. So the Win stat was almost completely based on timing, not contribution to the team’s win, while the pWin better reflects which pitcher contributed the most to the victory.

Royals vs. Tigers, August 5, 2015 (Tigers win 2-1)

Screen Shot 2016-04-08 at 11.10.51 AMIn Cueto’s second game for the Royals, the pLoss statistic runs into some problems. The Royals lost 2-1, and Cueto gave up both runs (both earned). Yet his WPA is higher than Kelvin Herrera’s, who gave up 0 runs in one inning of work. Herrera also didn’t allow any inherited runners to score (the “IS” column). So as my system is currently designed, Herrera gets the pLoss, which is obviously unfair. Let’s look at it a little closer and try to tweak the system.

In this case, the accumulation of WPA over 7 innings vs. only 1 inning for Herrera works in Cueto’s favor. And Cueto did pitch well – only giving up 2 runs in 7 innings of work. When it comes to determining pWins, I think this is a feature, not a bug – the pitcher who pitches the most in a victory should be the one most likely to get the pWin. But when it comes to a pLoss, this can penalize a reliever unfairly.

I think what stands out the most in this game is that Herrera didn’t allow any runs to score while he was on the mound – either inherited or his own. In no situation should a pitcher get a pLoss when that happens. So here’s my tweak:

The pLoss is given to the pitcher on the losing team with the lowest WPA and who allowed at least one run (earned, unearned, or inherited) to score.

In this game, that means Cueto.

A quick note about “inherited” runs. In the traditional system of scoring, the pitcher who allows a runner to get on base is fully credited with the Run if that runner scores, no matter if he is on the mound when it happens or not. I have always thought that is a bit unfair to the pitcher. But since allowing an inherited runner to score can impact your own WPA, it can impact whether you get the pWin or pLoss.

Royals vs. White Sox, September 29, 2015 (White Sox win 4-2)

Screen Shot 2016-04-11 at 11.01.05 AM

In this game, a 4-2 loss for the Royals, Cueto pitched 6 innings, giving up 3 runs. However, Joba Chamberlain gave up 1 run in 1 inning, and ended up with a lower WPA (-0.058 vs. -0.053) than Cueto. In the traditional system, Cueto gets the Loss; after all, he gave up three runs in a 4-2 loss – he should get the Loss, right?

I’m not so sure. After all, who did more to help their team win (and therefore prevent their team from losing) – the pitcher who pitched 6 innings of 3-run ball, or the one who gave up one run in one inning? The Royals were down 3-2 when Chamberlain entered the game in the bottom of the 8th, at which time they had an 87% chance of losing. But after facing just two batters, they now had a 96% chance of losing! There is a big difference that late in the game between a 4-2 deficit and a 3-2 deficit. Although I think timing is too influential in determining the traditional W/L stat, I don’t think timing is completely irrelevant either. Therefore I think it makes sense for Chamberlain to get the pLoss in this game instead of Cueto.

With the tweak to how pLosses are calculated, I think this WPA-based system is much stronger than the traditional one. However, this case study does reveal something about the pWin/pLoss statistic: it favors starting pitchers for pWins, and “favors” relievers for pLosses. I think this makes sense: someone who pitches more innings is helping his team more than someone who only pitches an inning (or even less), so favoritism towards the starters is justified.

In the final post of this series, I’ll sum up the pWin/pLoss statistic as well as look at some historical examples of it in use.