One of the data elements that is provided in the
MLB DFS master file is the home plate umpire for each game. I figured that this might lead to some interesting analysis – and maybe some of you have already begun doing so. I recently did some limited googling for DFS analysis related to umpires and found a few articles, but not much.
I did some very preliminary analysis on the data for the full 2015 regular season (which is available – for free - in the 2015 master file.) First, I selected all umpires who worked at least 20 games behind the plate during the 2015 season. There were 83 of them. The average number of home-plate games for each ump was 28, with a range from 20 games to 38 games.
For this analysis, I decided to look only at the performance of starting pitchers and starting hitters. I used DraftKings points (DKP) as the basis for the analysis, although I suspect similar results would be generated by FanDuel points or any other similar scoring system.
My objective was to evaluate how pitchers and hitters produced with specific umpires relative to their full-season averages. As a point of reference, the 2015 average DKP/game for starting hitters was 6.94, while the average starting pitcher produced 14.16 DKP/game.
For each of the 83 selected umpires, I calculated the average differential by which a player’s DKP for that game exceeded or fell short of his full-season average. These were tallied separately for hitters and pitchers. Hitters, for example, averaged 1.11 points more in games in which Manny Gonzalez was the home plate umpire. That is 16% above average. Similarly, starting pitchers averaged 3.26 points better than average when Toby Basner was calling balls and strikes, an outperformance of about 24%.
There may be factors that explain these variances other than home plate umpire. Ballpark influences, for example, might produce some of the variation, especially if umpiring crews are not geographically diversified over the course of a season. I glanced at the distribution of home plate umpires by ballpark, and did not see any striking disparities. No umpire worked the plate in more than 3 games at Coors field, for example, and 59 different umpires worked at least one game there. (Neither Manny Gonzalez nor Toby Basner worked the plate at Coors at all in 2015. At the other ballpark extreme, Gonzalez worked twice in scoring-hostile AT&T Park in San Francisco. But Basner didn't appear there either.)
I assembled a table showing the average +/- difference in DKP (actual points minus average) for each of the 83 umpires, expressing the differences both in points and as a percent of the average. Those tables are provided both as a spreadsheet and in PDF format. A few broad observations:
- In general, hitting and pitching results are negatively correlated. But there are some obvious anomalies.
- The most favorable umps for hitters were Manny Gonzalez and Ryan Blakney. As I noted earlier, there may be other explanatory factors at play for them. But the surface stats suggest that hitters produced more than 10% above their full-year averages when those two guys were calling balls and strikes.
- Gonzalez and Blakney were the two most hitter-friendly umps, but neither led the pitcher-unfriendly rankings – although they were both moderately unfavorable for pitchers.
- The most pitcher-friendly umps were Toby Basner, Mike Muchlinski, and Ben May. Starters produced more than 20% better than their averages in the presence of those three. Hitters in those games also underperformed by 10+%, ranking them near the top of hitter-unfriendliness.
- I did a little investigating into differences related to righty vs lefty starters, as it seemed plausible that umpires might exhibit some tendencies that are related to pitching arm. I did find some variation – although I couldn’t rule out that much of this result was driven by randomness due to the smaller sample sizes (i.e., most umps only called balls and strikes for 10-15 southpaw starters.) So I abandoned that analysis for now.
I’ll let you digest this data for now. If these results are credible and predictable (and I suspect some disparities are symptomatic of the umpires’ idiosyncrasies, but some are simply noise), then the next step is to be able to anticipate which umpires will be working the plate in upcoming games. That will be the topic of a follow-up article.