Fixing how wins are awarded in baseball

If it drives you bananas when a starting pitcher throws eight brilliant shutout innings and leaves with the lead but fails to get the win, then you’ll want to read this article.

If it really drives you mad when the reliever who blew that lead also gets the win, then you’ll definitely want to read this article.

In 2002 I found myself deciding to complain to Major League Baseball about the unintelligent way in which wins are awarded to pitchers. A way that often awards the win to the least deserving pitcher. But then I realized that would make me someone I don’t like – a complainer without a viable alternative to offer. So I resolved to come up with a viable alternative. But I couldn’t.

Then five years later, I could. And now, finally, eleven years after that, I offer you the merit win.

I’d rather it not be known as the “merit win”, but rather simply the new way of awarding wins, but until it becomes adopted as such, we’ll need a name to distinguish it from the current way. So, “merit wins”.

A key was to figure out how to give credit for innings pitched. Pitching 7 innings and giving up 1 run is probably more valuable than pitching 1 inning and giving up 0 runs. So I had to figure a way to give the correct amount of credit for those innings pitched.

The winning idea was that for each inning a pitcher throws, they get credit for the number of runs their own team scored per inning in that game. So if your team scored 4 runs and did not bat in the bottom of the ninth inning because they’d won the game, then they scored (4 runs)/(8 innings) = 0.5 runs per inning. Each pitcher is then credited with half a run per inning pitched. From this, the number of runs they allowed is subtracted. This gives the pitcher’s number of “runs ahead”. The pitcher with the highest number of runs ahead on the winning team earns the merit win.

It works for losses too – the pitcher with the lowest number of runs ahead for the losing team earns the merit loss.

One nice result is that you’re pretty much assured that the pitcher that earns a merit win will have a positive number of runs ahead, and the pitcher that earns the merit loss will have a negative number of runs ahead.

Example of how to calculate merit wins

Here’s a real-world example. On July 6 of this year, Jacob deGrom of the New York Mets threw 8 innings, giving up one run, in a home game against the Tampa Bay Rays, but was awarded no decision, leaving with the game tied 1-1. His teammate Jeurys Familia pitched a scoreless top of the 9th, then earned the win when the Mets hit a grand slam with two outs in the bottom of the ninth.

To determine who earns the merit win, first notice that the Mets scored 5 runs in 8 and 2/3 innings. That works out to 15/26 runs scored per inning, so we award runs to deGrom and Familia for their innings pitched at this rate. Here are their lines:

deGrom 8 IP, 1 R, 1 ER
Familia 1 IP, 0 R, 0 ER

So:

deGrom’s Runs Ahead = 8 IP \times \frac{15}{26} runs/IP – 1 R allowed = {4\frac{16}{26}} – 1 = {3\frac{8}{13}} runs ahead

Familia’s Runs ahead = 1 IP \times \frac{15}{26} runs/IP) – 0 R allowed = {\frac{15}{26}} runs ahead

deGrom has the higher runs ahead total, so deGrom earns the merit win.
There are two reasons why I chose this particular example. One is that it demonstrates one of the many ways in which regular wins are flawed and merit wins are not. The other is that it provides an example of a pitcher who may be denied his just reward – specifically the 2018 NL Cy Young award – because the current way of awarding wins and losses has made his undeserved poor record even worse than it needs to be.

A flaw of the win – dependence on which team bats first

In our real-world example above, what if the pitching performances, and each team’s turn at bat, had gone exactly as it did, the only difference being that it had been a road game instead of a home game for the Mets? Then deGrom would have left the game at the end of the bottom of the 8th inning, instead of leaving in the middle of the 8th inning as he did. Also, his team’s four-run scoring outburst would have occurred in the top of the 9th inning, instead of in the bottom of the ninth. deGrom would still have been the pitcher of record when his team took that lead for good in the top of the ninth, and because that’s how wins are currently determined, he would have earned the win.

Pitching 8 innings for the home team got him the benefit of only 8 innings of scoring by his team; pitching 8 innings for the visiting team would get him 9 innings of scoring by his team, increasing his odds of earning his team’s eventual win. In fact, no matter when the starting pitcher leaves the game, he always benefits from an extra inning of his team’s offense when pitching on the road. This frequently results in an arbitrary switch of which pitcher earns the win. I consider that a flaw – do you, too?

Merit wins are never arbitrary in this way. They don’t care about the order in which teams bat.

There are a lot of other ways in which the current way of awarding wins is flawed, and merit wins are not. One of the worst is described at the very beginning of this article. I expect to soon post a full description of all of them.

Justice for Jacob deGrom

The reason I’m finally getting this idea out now is that I hope it can save Jacob deGrom’s chance at winning the 2018 National League Cy Young award. It’s in jeapardy, despite his clearly being the best pitcher in the leage when you look at all statistics other than wins and losses. And the reason it’s in jeapardy is that his win-loss record is only 8-9, which is far from being a Cy-Young-worthy record.

Here is some of the talk about it.

But if you evaluate based on the merit win method, deGrom gains 3 wins he didn’t earn before, while losing none of the 8 he had. He also loses 4 of the losses he had before, while not gaining any of the 10 he didn’t have. Based on merit wins, his record becomes 11-5 – and that, I believe, should be good enough to convince Cy Young voters to vote for him.

(Early in the season, he left three games with a lead after 7 or more innings pitched, only to have the bullpen blow the lead and lose the game. Had it not been for those three blown leads, deGrom’s record would be 11-9 by the conventional method, and 14-5 by the merit win method.)

For merit wins to improve a pitcher’s record by this much over traditional wins is rare. With 3 wins added and 4 losses substracted, the merit win method improves deGrom’s Win-Loss differential by 7 games. In the 2012 season (the only season I’ve fully analyzed), no pitchers had a bigger improvement, only 5 had as much of an improvement, and only 4 improved by 6.

On the whole, in 2012, starting pitchers had 9.1% percent of their wins taken away using the merit wins method, but had new wins added totalling 18.3% of their regular wins total, for a net increase of 9.2% in their wins total. They had 13.6% of their losses taken away, and another 9.9% added for a net decrease of 3.6% in their losses total. So merit wins do tend to improve the records of starting pitchers, yet what is considered a “good record” going by regular wins and losses is likely to also be a good record when going by merit wins and losses; we maybe need to increase our win expectation by one win per starter.

Another look at it: 49.4% of starting pitchers’ decisions in 2012 were wins; 52.5% of their “merit decisions” were merit wins.

Tiebreakers

In 2012, 9.1% of merit win calculations resulted in a tie, requiring a tiebreaker, and 2.6% of merit loss decisions required a tiebreaker. The tiebreaking procedure is certainly something I’d like to hear some good debate about. What I came up with on my own involves repeating the calculation using earned runs as the first tiebreaker; most innings pitched as the second tiebreaker (reversing this to fewest in the case of evaluating for losses); fewest (most) batters faced after that; fewest (most) baserunners allowed (by hit, walk, or hit by pitch) after that; fewest (most) total bases allowed after that; and finally, the last pitcher to pitch. In the calculations I’ve cited here, I used this tiebreaking procedure with the exception of skipping the total bases allowed criterion, for lack of data on total bases allowed per pitcher.

One last telling statistic about merit wins

There is a lot of information about the effects of merit wins that I found when crunching the numbers on the 2012 season, that I plan to share in other posts. For now, I want to end with just one of those statistics, which I think gets to the essence of why I’d like to see official wins calculated in this way.

In 396 games in 2012, the starter threw six or more shutout innings in a game his team ended up winning. In 31 of these games, the starter did not earn the win. That’s 7.8% of these excellent starting performances that could have been awarded a win, but weren’t. By contrast, in all 396 of these games, the starter earned the merit win.

In that they attribute a team stat to an individual, wins and losses have always been flawed, and flawed they shall remain. But at least let’s start awarding them to the right player. That would make them a little less flawed.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s