Why Rays’ Drew Smyly Received $2.65 Million in Arbitration

The deadline for teams to exchange figures with their arbitration-eligible players can be a contentious time. Luckily for the Tampa Bay Rays, all their hard work proved to be enough this offseason as they avoided hearings with all eight of their eligibles. What were the factors that led them to the salaries to which they agreed with their players?

We discussed Alex Cobb, Kevin Jepsen, and John Jaso yesterday, and you can find those here. Right now, we will delve into how Drew Smyly ended up with $2.65 million (plus $50,00 in incentives) in his first go-around through arbitration as a Super Two.

Drew Smyly achieved solid results as a starter with the Detroit Tigers in 2012 and was dominant as a reliever in 2013, but 2014 was his best all-around season. Smyly went 9-10 with a 3.24 ERA, pitching to a 7.8 K/9, a 2.5 BB/9, and a 1.1 HR/9 in 153 innings pitched. He capped his year with a 1.70 ERA in 7 starts after the Rays acquired him in the David Price trade.

In the interest of making all of these arbitration posts a little bit different, I decided to compare Smyly to each “regular” Super Two starting pitcher from 2007 to 2013. The three pitchers I removed were Cole Hamels and Gio Gonzalez (who both signed extensions), and Tim Lincecum (who won back-to-back Cy Young awards). Here is how all of the other pitchers performed prior to being Super Two-eligible.

Admittedly, there is lot of information there, and I might as well clarify the “Last Yr bWAR (fWAR)” column. It records the number of wins above replacement each player recorded in their last year before being a Super Two, both according to Baseball-Reference and Fangraphs. I needed a quick way to summarize these pitchers’ last seasons before arbitration, so WAR was a good fit.

In any event, just by eyeballing the data, we can see a few clear trends. The arbitration system appears to value winning percentage, innings pitched, and ERA (at least for everyone but the primary relievers at the bottom) quite a bit. Smyly is especially interesting because he has the best ERA, ERA+, FIP, and K/9 in the entire table yet is just 13th out of 17 in innings pitched.

Going to more refined analysis, every statistic in the table but games, K/9, and K/BB are significant predictors, meaning that they explain the data better than chance alone. The best model we can get, though, is going to involve multiple predictors, and to get to that model, we draw upon the statistical concept of variable selection.

Our goal is to find the model with the best combination of simplicity (having as few variables as possible) and explaining the data well. As it turns out, after looking at every possible model including every possible subset of predictors, we end up getting two of the three statistics from our eyeball test.

Innings pitched is easily the best predictor of arbitration payouts for this group, explaining 62% of the variance in the payouts by itself. If we add FIP to our predictive model, we can get that up to 89% of the data. Finally, the best model we can get also adds ERA, which allows us to explain 91.1% of the data when combined with IP and FIP.

If we plug Smyly’s statistics into that model, we end up with an expected payout of $2.844 million, which is not far from what he got but still doesn’t explain everything. To make up some of that ground, we will use one factor we haven’t talked about yet, which is percentage of games pitched as a starter.

Smyly threw more innings than Jordan Zimmerman and Juan Nicasio, but both of them were full-time starters who didn’t pitch more because of injury. Smyly, meanwhile, threw just 241 innings as a starter, and his 87.1 relief innings are not as valued by the arbitration process. Once we plug in percentage of games pitched as a starter, we can get as close as $2.748 million.

Schedule