There has been a lot of talk recently about the lack of run support for Matt Harvey this season, based on observations by those among us that eagerly await his starts (in other words, nearly all of us who continue to have a pulse and electricity). Theories about why this is the case run from "the team feels more pressure to perform while the ace is on the mound" to "the team feels less pressure to perform when the ace is on the mound," and a bunch of things in between.
While I am in no position to judge what, if anything, the psychology of having your top starter on the mound might do to the supporting cast, I decided to look a little deeper into Harvey's run support to see whether observations of lack of run support are true, and if so to what extent they are true. Before going further, I want to make one thing clear: while a pitcher's wins and losses might mean a lot to some, I don't care about them so much. I care a lot more about runs scored during his starts and how that affects the ability of the team to win those starts. So I didn't look at how many runs were scored before he was relieved, because that would shift the focus to whether the team was helping to credit him with a win, rather than whether the team won when he started.
Overall, the eyes seem to be deceiving us. In MEH's 21 starts over the full season, the team has managed to put up a total of 83 runs, for an average of 3.95 runs a game. Overall, they are averaging 4.09 runs per game over the entire season to August 1, my arbitrary endpoint for this post because I was too lazy to input last night's scores into my spreadsheet (so hard to add a row, so hard). So Harvery is getting just 0.14 runs a game, or 3.4%, less run support than the team's starting staff overall.
But this observation obscures something else that gives some credit to the eyez: Omar Little's run support since May 1 has been significantly less than team average. What has the team done for him lately? Not much it turns out. The overall figure is buoyed by an average of 6 runs a game of run over his six starts in April, and since May 1 he has averaged 3.13 runs per game, as compared to a team average of 3.89. So while overall team production is down slightly, his run support per start is down 20.7% compared with his support in April, and sits at 19.5% below overall per game team production for May 1 through August 1. For some good, if unsurprising, news, the opposing teams in Harvey's starts have averaged 3.05 runs per game, about 31% less than the average opponent scoring over the entire season. So Harvey is good at his job.
There are some issues with my methodology, not the least of which is that my data entry skills might have been a bit sullied by a cocktail or two. We also have the question of a small number of games played in April. I also didn't pull out Harvey starts from overall averages so the spread between Harvey starts and non-Harvey starts is a little wider than mentioned. For example, in non-Harvey starts over the entire season the team is averaging 4.22 runs per game, which means that compared to all other starters he is receiving 6.3% fewer runs. Still, this is not that huge a gap, especially compared to the numbers after the first month of the season, and is a lot less than what anecdotal evidence might have us believe.
Overall, it's fairly clear that while Harvey has received just a bit less run support than his peers, the gap has widened significantly since April. Perhaps this is a case of regression to the mean and the law of large numbers, or perhaps it is indeed a matter of psychology once the team developed increased confidence in him, or perhaps it is because of the angering of the baseball gods who are envious of his beauty and are vengeful, or whatever. I decided to look a little deeper to see if there might be an obvious reason.
Because he was not the starter at the opener, I discounted the idea that he's going up against the opposing team's ace very often due to scheduling, but in the spirit of captaining the mighty ship Obvious I wanted to take a look at opposing starting pitchers. I ignored relievers because it would take forever to figure it all out and generally opposing relief staffs are likely to even out over the course of many games. With a little searching around I found that indeed the discrepancy might not be an issue of the Mets bats going cold, but instead a reflection of the pitchers they were facing. I used fWAR to get a general sketch of opposing pitcher quality, fully expecting someone to come on here and complain that it's imperfect for this kind of comparison due to limitations based on injury, being sent down etc.
Opposing team's starters who faced Harvey in April have an average 2013 fWAR of 0.43, with 3 opposing starters having negative fWAR for the season (which kind of makes fWAR look good in this context). Looking at the month game by game shows that very often the Mets got to these pitchers early, and were left to face long relievers. This includes the game that the Mets scored 7 runs in a Strasburg start, and a Jose Fernandez start in which the Mets only managed 3 (Jose seriously inflated average fWAR for April, which tells you how bad the numbers were for the non-Strasburg others). From May until August 1, opposing starters have an average 2013 fWAR of 1.55, with none in negative numbers.
While it may be tempting to credit Harvey's apparent lack of run support over the season to hitter expectation or some other sort of team psychology, in fact in appears that Harvey's run support has been close to what the other starters are collectively receiving, if a hair below. Furthermore, while there does appear to be a strong split between what the team provided for him in April and what the team has done since, this looks like it might be the result of facing much better starting pitching since an anomalous April filled with scrubs and Strasburg laying an egg. This also supports the idea of regression to the mean over a long season, with the team coming back to its expected production once a more accurate range of opposing pitching is taken into account.
One counter-argument is that run support was inflated in April because of the atrocious nature of the opposing starters, and once the opposing starters got more in step the run support gap quickly widened. This might give some credit to our collective observations. However, I'm more inclined to believe that while you're not going to face a negative fWAR starter very often, it happens from time to time and in his starts it just happened to be clustered towards the beginning of the season. Therefore, it seems like the lack of run support this season is a myth, with an important, if imperfect, caveat.