Do Doubleheaders Impact Run Scoring?

This is a question I’ve asked myself several times this year when preparing for daily fantasy contests. My assumption was that run scoring in doubleheader games might be lower because of fatigue and because it’s not uncommon for teams to start bench players in one of the two doubleheader games. But after a little digging, I found that assumption to be incorrect.

The only data I could find on doubleheaders was a list of doubleheaders starting in 2002 on ESPN.com. From 2002 to 2013, the average number of runs scored in a major league game was 9.13. There were 296 doubleheaders played in that time frame. To my surprise, the average number of runs scored in the doubleheader games was higher. The average number of runs scored in Game 1 of doubleheaders was 9.35, and the average number of runs scored in Game 2 was 9.26.

When I hypothesized the effect of doubleheaders on run scoring, I assumed the use of lesser hitters would reduce run scoring. I’m guessing the reason that hypothesis was wrong is that the use of lesser starting pitchers has a bigger impact. It would be far too big a task for me to go back and look at all the starters in doubleheaders and compare their numbers to that of the average starting pitcher. But we’re all aware that call-ups and long relief men get spot starts frequently in doubleheader games. And, again, I don’t have the time to go back and look at all the starters, but it’s my perception that the pitcher making the unexpected spot start tends to pitch Game 1 of the doubleheader more often than Game 2. There could be other reasons more runs are scored in doubleheader games (for example, a team’s worst relievers are probably going to have to appear), but poorer starting pitching seems the most likely reason to me. If you have other ideas, feel free to suggest them.

Of course another reason for the difference in run scoring in doubleheader games could simply be variance. Over 29,000 games were played between 2002 and 2013, and I’m telling you that the results of a specific 592 game sample out of those 29,000 games were different because the quality of pitching wasn’t as good as it normally is as opposed to this just being random. I know there are ways to test the significance of this, but the methods by which such a thing would be tested are beyond my capabilities. I spent several hours in Excel trying to determine the significance of this, but I’m anything but certain that the results of my efforts were correct. I’ve got an email in to Steve Staude who should be able to answer this question for me. I’ll make an appropriate update to this post if necessary after Steve looks into it.

I also want to point out that my other inclination that fatigue might reduce run scoring was not a thought worth having. It’s not possible to measure and test fatigue in that manner. That was an easy narrative to apply, and one that proved false. As most narratives do. Although I’m quite certain the Rangers will lose Game 2 of a doubleheader someday soon by a score of 1-0, and I’ll get a text from my father, the king of all narratives, saying the Rangers were just too tired in that second game.

One other interesting thing I found when looking into this issue was that a disproportionate number of the highest team run totals in a single game occurred in a doubleheader game. This is somewhat arbitrary, but teams scored 17 runs or more in a single game 154 times from 2002-2013. Of those 154 instances, ten occurred in a doubleheader game. So in the last 12 years only 2% of all games were part of a doubleheader but 6.5% of the 154 highest team run totals occurred in doubleheader games? Seems fishy. But I did absolutely zero looking into the statistical significance of this. It could be nothing.

My guess is that I’ll get some sort of confirmation from Steve that there is something to this. And, again, if it turns out not to be significant, I’ll update this post. But for now let’s assume it does mean something. How can we apply that to DFS? I’m a big fan of stacking, so on days with doubleheader games I’ll be looking to invest more in stacking players from teams facing a below average spot starter (revolutionary analysis, I know). Obviously you’ll have to consider other factors like weather, platoon splits, etc. and avoid that particular stack if there’s something offsetting the upside. But in GPP contests especially, it might not be a bad idea to lean towards stacks against lesser pitchers in doubleheader games. And if you were trying to choose between two players or two different stacks and one was involved in a doubleheader game against a spot starter, using the doubleheader as a tie breaker is probably a good call.