I’ll get right to the point. A lot of people think Sonny Gray needs a personal catcher. His season, thus far, has been inconsistent. He looked good in his first start, save for throwing too many pitches (probably would have pitched into the 5th inning if it wasn’t his first start of the year), and his second start really wasn’t that bad. These last two, however, have been markedly worse. The common theme for these four games so far: the starting catcher.
Austin Romine was behind the dish for the first two, while Gary Sanchez caught Gray in the latter two. The idea of Romine being Sonny’s guy dates back to last season. This year, the Yankees sluggish start seems to have accelerated these talks. But is there actually any credence to the personal catcher idea, or is it just an overreaction to two bad early outings? In this article, I’ll explain the results of four regression models I ran on Sonny Gray’s ERA, BB/9, H/9, and K/9 based on who his catcher is, and then summarize what this means and the merits of having a personal catcher. Data used for this is from his entire career as a Yankee.
For this regression, the intercept represents the projected outcome when Austin Romine catches Sonny Gray against a team whose record is at or below .500. I included the opponent’s record in the model as a sort of explanatory variable which helps with endogeneity (fancy statistics term for variable biasedness), and heteroskedasticity (another fancy term for something you don’t need to worry about). For each model, the coefficients represent units of deviation from the intercept – AKA, how it compares to when Romine catches against a team with a losing record. R^2 shows how much the model predicts the outcome. I’ll let you all look up interpretations on that if you’re interested. Additionally, if you have further questions on the math behind this, feel free to DM me on Twitter, or do a simple Google search on linear regressions.
|Opponent Above .500||3.32*|
|* denotes 5% significance level
R^2 = 31.47%
In the ERA model, all variables are statistically significant, which means they matter. Against a winning team, Gray’s ERA is projected to increase by 3.32 points. When Gary Sanchez catches, his ERA is projected to increase by a whopping 5.07 points. So, the total interpretation for that would be “When Sonny Gray pitches to Gary Sanchez against a team with a winning record, his ERA will be 3.32 + 5.07 (8.39) points higher than when he pitches to Austin Romine against a team with a losing record.” This style of interpretation can be followed with any combination of variables for any of the models.
This simple model certainly gives credence to those arguing for a personal catcher. Gary’s presence behind the plate does inflate Gray’s ERA, and, statistically speaking, IT MATTERS!
Now, I mentioned the term endogeneity earlier. It essentially means that the variables could be correlated with unobserved variables in something called the “error term.” These variables in the error term represent anything that might affect Sonny Gray’s ERA which is not represented in the model; i.e. weather, specific batter matchups, control of his pitches, spin rate, and even intangibles like his mood. However, this should not come into play here because Gary Sanchez catching has nothing to do with most of these unobserved variables. A potential argument you could make is that Gary puts Sonny in a bad mood which affects performance, but it is literally impossible to quantify and should be captured in the Sanchez variable anyways. All in all, it is impossible to argue against the fact that Sonny Gray pitches worse when Gary Sanchez catches and it statistically matters.
This doesn’t necessarily mean I think Romine should catch all his starts, but I’ll have more on that later. For the next three models, I will have just a few comments to make and then get to my final analysis.
|Opponent Above .500||1.32|
|R^2 = 9.22%|
None of the variables in this model are statistically significant from zero. This means that neither the catcher nor the opponent’s record influence the amount of walks per nine innings Sonny Gray surrenders.
|Opponent Above .500||4.81**|
|* denotes 10% significance level
** denotes 5% significance level
R^2 = 30.30%
This model shows the same trend as ERA. The intercept is not significant at the 5% level, but for our purposes, I accepted a larger significance for a better analysis. When Gary catches, or Sonny pitches against a better team, he allows more hits. In regards to Sanchez, it’s hard to speculate on why this could be, but he is probably leaving more pitches over the plate. Just a hunch (I based this on his pitch location heat maps on fangraphs).
|Opponent Above .500||0.04*|
|* denotes 5% significance level
R^2 = 17.62%
Based on the other three models, this one makes sense. When Gary catches, Sonny Gray strikes out almost four less batters than when Romine catches. Although the team’s record is significant, it barely influences the projected outcome, so strikeouts fall almost entirely on the catcher (within our model, that is).
So, we know without a doubt that Sonny Gray does better when Austin Romine catches. But does this mean he needs the personal catcher tag? It’s a tricky subject, and people on both sides of the argument should understand the others’ point of view.
Those who are pro-personal catcher have an easy argument: stats don’t lie. But that isn’t the whole story. Gary Sanchez is the best offensive catcher in the league. Not one of the best, he is the best. Therefore, he needs to be in the lineup as much as possible. On days of rest from catching, he should be DH-ing. Unfortunately, when El Gary DHs, someone else has to leave the lineup. The top candidate has been Gardner, but this could change (I can see Hicks getting the boot). Not only is Gardner the catalyst, both players I mentioned are better outfielders than Giancarlo Stanton, who would need to play left field when Sanchez is in the DH slot. Not to mention, Austin Romine is a below average hitter. Overall, the lineup’s offensive and defensive abilities get worse.
Days when Gary can DH are not the biggest problem, however. What happens when Sonny Gray pitches a day after Sanchez needs a day of rest? Does Romine now have to catch two or three times per week? And if we make the playoffs (sorry for the lack of enthusiasm, but I REFUSE to jinx the team), is Boone required to sacrifice offensive production because of Sonny Gray? Personally, I am not a fan of these scenarios.
“How have personal catchers worked in the past?” you might ask, or “when have they been implemented?” The obvious situation is knuckleballers, like Tim Wakefield or R.A. Dickey. A manager has no choice there. Another exception would be if Greg Maddux or Jon Lester is on your team. Sonny Gray is good, but he is not Greg Maddux, nor is he prime-Jon Lester. The Yankees had a recent example with A.J. Burnett and Jose Molina. Burnett and Jorge Posada did not get along, so Girardi made a call. There was a lot of “mystery” surrounding this whole scenario, but everyone could read between the lines. Sometimes, personal catchers are necessary, but sometimes it’s a stretch.
The anti-personal catcher crowd will tell you Sonny Gray is mentally weak and just needs to figure it out, or if Romine is doing something better, then Sanchez needs to find out what that is. Both are potentially true, but there is one thing this whole argument comes down to: runs scored vs. runs allowed.
Aaron Boone’s decision needs to be based on how many runs are scored when Romine is in the lineup, and how many runs are allowed when he catches. That’s it. (I plan on doing an analysis of the team’s runs scored vs. given up when Romine catches in a later article). If those two metrics cross each other, then Gary needs to catch. If however, Romine catching moves those metrics further apart, is it so crazy for Sonny Gray to have a personal catcher? A manager’s goal should be to maximize his team’s chances to win on a certain day. If Boone is not doing that for the sake of what is “right,” then he is not taking his job seriously.
To please both parties in this argument, here is my solution: no personal catcher for Sonny Gray, but if it works out, start Austin Romine at catcher. Now, “if it works out” has room for interpretation, but basically Boone shouldn’t be afraid to play Romine on Sonny Gray start days. I’ll ask this question again: if you know Sonny pitches better to Romine, why would you go out of your way to hinder his performance? Usually, I believe that compromise does not reach the best outcome, but in this case, compromise may be king.
**Stats from MLB.com