The effect of umpires seems to be a contentious subject between baseball bettors and fantasy players alike.  Some believe it is vital while others believe it’s not worth considering.  The correct response is likely somewhere in the middle but let’s take a look at data before we let our assumptions get in the way.

For this post, I looked at 3-year samples of umpire data starting in 2005 for umpires that were good “Over” umpires and good “Under” umpires.  For an umpire to be counted, they were required to have had at least 30 games behind home plate and a return on investment (ROI) of at least 10% for the 3-year period.

I then looked at the results of the same group of umpires for the next season (n+1) and the next three seasons (n+3).  Here are the results for “Over” umpires:

SeasonsResults (Units, ROI)n+1 Resultsn+3 Results
2005-2007+107.2, 17.3%+2.36, 1.2%-38.12, -7.3%
2006-2008+73.71, 18.2%-11.25, -9.1%+10.75, 3.0%
2007-2009+84.92, 14.2%+13.22, 6.4%+7.09, 1.1%
2008-2010+100.99, 18.4%-1.86, -1.2%-16.82, -3.9%
2009-2011+106.21, 13.5%-11.0, -4.5%-59.86, -9.2%
2010-2012+141.21, 15.6%-42.75, -13.4%-29.59, -4.5%
2011-2013+127.01, 15.4%-2.77, -1.3%N/A

And here are the results for the “Under” umpires:

SeasonsResults (Units, ROI)n+1 Resultsn+3 Results
2005-2007+70.44, 14.4%+6.53, 3.7%+7.61, 1.4%
2006-2008+83.17, 14.6%+17.91, 8.4%-8.27, -1.4%
2007-2009+131.19, 18.5%-4.57, -1.6%-20.14, -2.4%
2008-2010+168.37, 14.8%+12.2, 3.2%+63.32, 5.7%
2009-2011+164.66, 17.9%-15.12, -4.9%-0.74, -0.1%
2010-2012+163.56, 14.6%-11.58, -2.9%-16.76, -2.0%
2011-2013+188.41, 16.9%+16.25, 5.0%N/A

Using previous over/under results to determine future over/under results has not been a profitable endeavor.  This isn’t to say that umpires don’t affect the games at all, as they all have differing strike zones which will alter each game.

Since 2007, the best “Under” umpire has been Bill Miller (156-104, +40.86 units won).  Via a website called let’s take a look at how often Miller calls strikes compared to league average (click to enlarge):

Miller calls both the high strike and the low strike more than the average umpire.  This appears to be the root cause for why so many of Miller’s games have gone under the total.  Just as a good framing catcher can “steal” strikes for his pitcher, an umpire forcing hitters to expand their strike zone will positively impact the pitchers and in turn cause a game to score fewer runs.

While the tables above suggest to ignore previous results, we can look deeper into each umpire’s strikezone to see if there is an underlying reason that helps explain the extreme outliers.  Many umpires will skew over or under based on randomness alone, especially over small sample sizes.  But with all of the information available to us today, you can differentiate between the predictive data and the random noise and apply it to your betting strategies accordingly.