Trevor Bauer has been good during the day, but terrible at night. Does that mean anything?

Trevor Bauer has had a trying season. He was doing reasonably well during the first half, posting a 4.02 ERA with a 3.97 FIP, but in the second half he came apart, with a 5.33 ERA and a 5.04 FIP. There's another split for which the difference in Bauer's numbers is even more dramatic.

During ten day games, Bauer has posted a 3.52 ERA, but during twenty appearances at night, Bauer has a 5.35 ERA. We might just chalk that up to small-sample noise, but his career numbers for these splits are even more dramatic, with a 3.39 ERA in day games, and a 5.32 ERA in night games.

When this was pointed out in the comments of a thread last week, someone asked if there was anything to them or not. At least one person quickly jumped in to say there was nothing. There were probably lots of pitchers with differences like that, and even if there weren't, there wasn't anything significant about it. While if forced to guess, I'd have guessed the same thing, I responded that I thought there was enough to be worth investigating, and that I'd do just that. I dislike it when people simply assume their gut reaction is correct, without bothering to check. It's lazy, and it promotes ignorance. Even if your gut is right more often than not, when it's possible, you should go further.

Baseball-Reference has day/night splits, and is the source for all the numbers below, and all information is current through September 29, 2015.

The first thing I went to check on is which pitchers have the largest difference between their day and night split so far this decade. I wanted a larger sample of innings pitched to work with. When I set the limit at a minimum of 200 innings for each split, there were 111 pitchers who qualified. Among that group, there are 12 who have a difference of at least one full run in their ERA.

A negative number in the "Difference" column means the pitcher's daytime ERA was lower; a positive number means the pitcher's nighttime ERA was lower.

Better during the day than night:

Better during at night than during the day:

In case you're wondering, we shouldn't expect more pitchers to do well during the day, not based on league averages anyway. From 2011-2014, the MLB-wide ERA during the day was 3.90. During the night, it was 3.90. If there's anything to pitching being easier during one time of day than another, it doesn't show up in ERA.

Differences like those we see from Kazmir and Lynn are interesting, but may not mean anything. My next step was to find pitchers who'd had an especially large split difference for a couple seasons, and then see what happened the following year. Did the trend continue for them, or did the next season seem to provide a random set of results.

The trouble with this was that the only way I knew how to do it was to open up the day and night splits for two-season stretches next to each other, and then go back and forth to find the same pitcher on each list. It was very time consuming. I looked at four two-year stretches: 2010-11, 2011-12, 2012-13, and 2013-14. I also limited it to pitchers with at least 100 innings in each split for those two years. Each of the four groups provided roughly 75 pitchers to cross-check, and I was looking for pitchers whose day and night ERA figures were at least 1.50 runs apart.

Next, I looked at each pitcher in the following season, but only kept track of the data if the pitcher had at least 50 innings in each split for that next season. That's a fairly small sample, but I thought it would be enough to at least get a sense of whether there's anything predictive about the splits.

For better or for worse, only 18 three-season stretches met all of those criteria, 10 in which the pitcher much better during the day in the first two seasons, 8 in which the pitcher was much better at night.

Better at day:

Of the ten pitchers who had a significantly lower daytime ERA during the first two years, seven of them also had a lower daytime ERA in the third season, and the four pitchers with the largest difference in either direction for the third season were all from the group that continued to perform better during the day. Those two factors seem like evidence that there is something to the split difference, but the evidence can't be described as anything overwhelming.

Better at night:

Of the eight pitchers who had a significantly lower nighttime ERA during the first two years, five of them also had a lower nighttime ERA in the third season, but two of the three largest differences for the third season swung in the opposite direction, meaning the pitcher suddenly performed much better during the day. Jeff Samardzija was much better at night in 2011, 2012, 2013, and 2014, but has now been dramatically better during the day this season. (More accurately, he's been bad during the day this season, but really bad at night.)

Of the 18 data points from either of the two charts above, 12 of them did show a continuation in the third season of the trend from the first two seasons, and five of the seven biggest third-year differences were a continuation. I think that's enough to count as support for the idea that there is something to day/night splits for those on the extreme ends of things, but I wouldn't characterize it as strong evidence.

A study that looks at a larger pool of players would be helpful, but somewhat unwieldy. I could perhaps have looked at the same two-year outliers, but followed up with another two-year follow up, rather than just the single season. I could have included players with less extreme splits for the first two years (perhaps anything larger than 1.00, instead of 1.50.) I doubt there's a major reveal to be found through a bigger study, but I'm also not willing to entirely dismiss the notion that some pitchers really do perform better during either day or night games.

In the meantime, Trevor Bauer will get one final start this season, and it will be today. Wait, make that, tonight.