At the beginning of the 2015 season, the folks over at Baseball Prospectus introduced a new run average stat for pitchers called deserved run average, or DRA. It's an attempt to take the next step in the work FIP started by controlling for only that which the pitcher himself has influence over. And it has some interesting things to say about Tigers' ace Justin Verlander.
When sabermetrician Tom Tango developed fielding independent pitching (FIP), he attempted to control for the vagaries of the defense behind the pitcher by considering only strikeouts, walks, and home runs. Everything that happened when a ball was put into play was thought to be the result of luck or the defense. By trimming out everything that appeared to be out of the pitcher's control, Tango sought to create a more accurate statistic to judge pitchers free from the effects of bad defense, varying ballparks, game time temperature, and other factors.
At that point in the development of advanced metrics, defensive statistics were still in their infancy. Much progress has been made since that point. While ultimate zone rating (UZR) or defensive runs saved (DRS) aren't perfect, they represent a huge advance from the days when errors or fielding percentage were all we had to go on.
DRA incorporates a team's defense into its calculation, but it goes well beyond that alone. From Baseball Prospectus:
DRA controls for the context in which each event of a game occurred, thereby allowing a more accurate prediction of pitcher responsibility, particularly in smaller samples. DRA goes well beyond strikeouts, walks, hit batsman, and home runs, and considers all available batting events.
But where DRA really breaks new ground is in its emphasis on run expectancy. By adjusting a pitcher's run average by the run expectancy for any given scenario, DRA more accurately determines what the pitcher deserves to be charged with. For example, if a starting pitcher is pulled from the game with a man on first and one out, and the pitcher who relieves him allows that runner to score, DRA would only charge the starter for the .562 runs that represents the average outcome for a runner in that base-outs situation.
This makes sense, right? Why should the original pitcher be penalized the same amount for leaving with a runner on first base with two outs as he would for leaving one on third base and no outs? The situation he left his reliever with should be factored into how badly he is penalized for that runner scoring.
Of particular interest to Tigers fans is their team's ace, Justin Verlander. DRA provides even more compelling evidence that Verlander found his way back into the company of the best starters in the game last year. He posted the fourth-best DRA of any starting pitcher in the game last season. Only Clayton Kershaw, Zack Greinke and Jake Arrieta posted a better mark in 2015 among starting pitchers.
DRA adjusts for several factors that affected Verlander's ERA last season. Understanding how DRA uses run expectancy makes it obvious how a terrible bullpen can allow runs that Verlander was fully charged for. It also corrects for the Tigers' lousy pitch framing work last season. Further, it controls for the differences in ballparks, the umpire strike zone variations, and a host of other minor factors. Verlander also gets credit for being very tough to steal bases against. All these little details have a value, and in DRA, they produce a more honest expression of how well the pitcher, by himself, prevented runs.
As a new statistic, DRA is understandably going to take some scrutiny before it becomes a more popular method for evaluating pitcher performance. Its complexity naturally makes it a target for skepticism. Yet, by controlling for a lot of these less obvious factors, it actually builds into its method many of those intangibles which skeptics of advanced metrics like to point to as blind spots in advanced statistics.
A much more detailed explanation of DRA is available here.