Assigning Pitching + Position Player Wins

Last Thursday, Ben wrote the above piece, talking about how there might be a need to revise the measure by which pitching wins are handed out. One of the ideas in the comments - okay, it was mine! - was to use Win Probability to decide who was the team’s best (or worst, in the case of losses) pitcher. This would remove the need for the starting pitcher to go an arbitrary five innings to get the W, and simply award the victory to the pitcher who had the most impact on the game’s final score. This seems fair, because there have been times a player has fallen short of five, but has still been the most significant pitcher in the game.

Take the game on June 1 against the Mets, for example. Slade Cecconi started for the D-backs, and pitched very well. He allowed five hits and no walks, leading to just one run allowed. But because he was lifted with two outs in the fifth, even though the D-backs were 4-1 up and cruised to an easy 10-5 victory thereafter, he got a no-decision. The W went to Kevin Ginkel instead, who threw eight pitches and had a +4.9% Win Percentage. That’s weak sauce compared to our starting pitcher, who threw seventy-one pitches and had a +12.2% Win Percentage. Nobody else reached even +1%. To be honest, the W should have gone to the hitters that day. But if any pitcher deserved a W, it was Cecconi.

There are other circumstances where applying the rules can seem unfair. Reliever A comes into a tied game in the seventh, works a clean inning but then gets lifted after walking a man with two outs in the eighth. His replacement, Reliever B, allows a home-run. But it’s the man on base which represents the go-ahead run, and that’s Reliever A’s responsibility. So he would get the loss in this situation. But there’s no doubt about the most infamous non-win in team history. That would be Randy Johnson’s start on May 8th, 2001, when he struck out twenty over nine innings, and got a no decision. Though by Win Probability, Miguel Batista’s nine shutout innings on May 28th 2002 (a 1-0 win in 18 innings) rates higher, at +66.6%

To address this, I went through all of the Diamondbacks’ games this year, and redistributed W’s and L’s, based purely on the pitcher with the best and worst Win Probability. This works a little bit like Fangraphs Shutdown/Meltdown stat, with a couple of changes. Firstly, only one W or L can be given per game; there can be multiple SDs and MDs in a single game. Secondly, both starter and relief pitchers are eligible for the decision, SD/MD is purely a bullpen stat. Thirdly, there’s no minimum limit to qualify. You can get the L as the “worst” pitcher in a 1-0 loss, despite positive WP, as in this game last year. Sucks to be Luis Frias in that case, especially as he only faced one batter and retired him. No stat is perfect. :)

That all said, the following table lists the redistributed wins and losses for pitchers.

This aligns quite well in terms of what we’ve seen from our pitchers this year. Paul Sewald was great, until a week ago, when he starting vacuuming up L’s like a Roomba on meth. Zac Gallen has generally been good, and Martinez seems deserving of higher-leverage action. At the other end, it’s mostly as you’d perhaps expect too. The back end of the bullpen, the revolving door which has been the fifth spot in the rotation, and Jordan Montgomery, have all been responsible for more losses than wins. This seems much more accurate than traditional W/L, which has Montgomery and Gallen both 6-5. Ryne Nelson’s mark (regular 6-6) is very close, but Kevin Ginkel (6-1) looks a lot less impressive by this metric.

Wins for position players

Why should pitchers get all the fun? Why is it only the men on the mound that get the letter - good or bad - by their name in the box-score, while the guys at the plate get to skate all responsibility? Let’s see if we can rectify this in the same way, coming up with a totally new metric for position player wins and losses. You could make the case that, for any game, there should be only one decision, which can go to either a pitcher or a hitter, depending on who has the most impactful Win Probability. For there are games where pitching is the decisive factor, and others where it’s hitting. You could perhaps look at the total WP for each, to decide who gets the decision, then award it to the best/worst in that area.

However, life’s too short for that. So I just did it the same way for hitters, so every game will have one pitcher and one hitter who gets a decision. For batters, Win Probability is typically going to be more spread out, because in any game there will always be at least nine of them. The average game this season has used only 4.2 pitchers. But we find a larger number of different pitchers have been given decisions for Arizona: 21, compared to only 16 hitters. However, the D-backs have simply used more pitchers this year: 27 (discounting position players mop-up), compared to only 18 hitters. Though one of the “hitters” to get an L was actually a pitcher, even in the age of the DH. I’m sure you can guess who that was!

Here’s your chart for position players:

I’m kinda impressed Christian Walker has been the most impactful hitter in almost twenty percent of all Arizona’s games, and nearly one-quarter of their victories. Joc Pederson’s ratio of wins to losses is even better, though Ketel Marte has more of both to his name. At the other end, no surprise to see Corbin Carroll and Eugenio Suarez in the bottom three, but I wasn’t expecting Lourdes there. However, looking at the totals for Win Expectancy on the team, he is solidly in negative territory. Alek Thomas though: eleven games and already -62.7%. Here’s a stat: Thomas has not had a positive Win Probability in any game this season. Include last year, and it’s 15 consecutive regular season game with a PA, in negative WP.

I’ll dig more into that for tonight’s Gameday Thread, I think. Anyway, I’ll keep an eye on these numbers and do an update on them in a month or so. If I remember!