November 17, 2015
So, Hey, What if the Mets Had Intentionally Walked Wade Davis
During Game Four of the World Series, the Royals used their patented late-inning devil magic to come back and take a lead against the New York Mets. This turn of plot was old hat to the Royals, who outscored opponents 18-0 in the ninth inning (or later) of postseason games in 2015, though by their standards the twist came surprisingly early—during the eighth inning. That meant that the Royals’ closer Wade Davis was forced to approach the plate as a hitter for the first time since 2013.
The result wasn’t good, but the reality was even more comical than what the box score shows:
Davis gleefully jogged back to the dugout having watched four mid-90s fastballs cross home plate against him. This of course, made me wonder…
What if the Mets had just walked him intentionally?
Seriously, what if? Davis would have to then spend some time on the basepaths, instead of in the dugout with his arm wrapped in a towel. It was 51 degrees when first pitch was thrown, so Davis would be standing out on the bases at 11:30 with temperatures several degrees lower. It might have messed with his pre-pitching routine if he was stranded to end the inning and he had to jog into the dugout or wait for his glove to be brought out by one of the Royals players.
In theory, walking Wade Davis isn’t necessarily the worst idea in the world. The Mets were already down two runs, already heavily disadvantaged against a pitcher who simply doesn't give up two runs. They would slightly increase the chances of going down three (or more), but perhaps throw off Davis' mindset and increase (however slightly) the chances that he has an uncharacteristic meltdown. We can easily measure how much risk it would be to walk Davis; we might also be able to identify whether there's a potential benefit.
The route the Mets took, the gifted strikeout of Davis, left the Royals with no runners on and two outs in the inning. In 2015 teams scored 0.1 runs per inning in those circumstances. The alternative, walking Davis would leave the Royals with a runner on first and one out, carried a run expectancy of 0.5 runs in 2015, meaning that the prospect of walking Davis would generate 0.4 more expected runs than striking him out.
While Baseball Prospectus’ run expectancy table is built on aggregate numbers across MLB each season, it’s worth noting that Davis was followed by the top of the Royals lineup, so it’s not as if the batters following Davis were easy outs.
Now that we know the bar that Davis would need to surpass, we can consider whether or not walking Davis would have actually helped the Mets. The next step was to pull some similar data to understand whether walking opposing relievers produces lackluster results in subsequent plate appearances. To do that the BP Stats team helped pull data on every relief pitcher plate appearance dating back to the 2005 season. In total there were 511 relief pitchers who threw both before and after a plate appearance where they reached base, making up the sample for this exercise. It’s worth noting that the average reliever in this sample was on base for just over seven pitches, but the number ranged from just one pitch to as many as 25.
In order to compare pitchers in both scenarios I took the average of their performances before and after the plate appearance in which they reached base. The denominator for most of these analyses is plate appearances because relievers often don’t work full innings.
Before their plate appearance, the relief pitchers in this study posted the following stats:
Overall, the pitchers in the study gave up an on base percentage-equivalent of .289, with strikeouts in 20 percent of plate appearances. After their plate appearances the numbers changed a fair amount:
The OBP-equivalent of opposing hitters went up nearly 30 points, and home runs per plate appearance increased one percentage point. Strikeouts also came down 10 percent. It's not an insignificant difference—except, of course, it is, because we don't know whether the variable we're considering (reaching base) was a factor at all. Quite likely pitchers are just worse in their second innings of relief, even when they're pitching well enough that their managers allow them to bat.
Sooooo, let's look at our control group: Pitchers who had a plate appearance where they did not reach base, as Wade Davis did, followed by another appearance on the mound. During the same timeframe there were over 2,700 relievers in our sample. Their before and after numbers, compared to our test group, look like so:
The numbers are similar. The effect of sitting out an inning, and possibly also having a plate appearance has a big effect; yay for the Mets! But the simple act of reaching base is of little to no value in general. Or, taken at face value, negative value (though that seems unlikely).
Now, as mentioned above: It was real cold. The average temperature at first pitch in all games we looked at was 77 degrees, more than 25 degrees warmer than New York on the night of Game Four. If we restrict our sample to games that started at 55 degrees or below (a sample that is tiny, but we're desperate here), the results are as follows:
Okay, the best remaining case for walking Davis: This type of analysis looks at the typical results of hundreds of events, stretching across nearly a dozen seasons. But Davis and the Royals weren't the collective average of 571 attempts; they were just one, and maybe that just once would have been the time it mattered. Maybe they were atypical. There's no way of knowing, but it's no less sensible than the Royals believing Alcides Escobar could win the game just by swinging at the first pitch.
Of course, that's not very sensible, either.
The outcome for the Mets might have been very different had they just let Wade Davis take first base. There’s really no logical reason to think that of course, but at that point in the World Series irrationality was about all the Mets had. They were desperate for something to matter. Walking Wade Davis would have been really odd, pretty cool, and certainly entertaining. It just probably wouldn’t have helped the Mets at all.
Special thanks to Rob McQuown and Matt Perez for data and calculation assistance.