Deserved Run Average: An attempt at quantifying pitcher’s true performance

For as long as baseball can remember, pitchers have been judged on the amount of runs they’ve surrendered during an outing. It’s the barometer by which fans have evaluated players and teams have delivered contracts.

In many ways, we still do. During conventional games, the first statistic viewers see is ERA: the average amount of earned runs a pitcher gives up over the course of nine innings (one game).

But ERA is inherently flawed. It tells us what happened without providing a reliable indicator for what will happen. Sure, if someone has a 3.44 ERA over 12 career seasons and has a 2.32 ERA this season, it’s conceivable that regression is ahead. But there is no real way to evaluate what he’s currently doing.

Not only that, ERA relies heavily on fielding, bounces, scoring effects, and a plethora of other factors that really don’t have anything to do with how good a pitcher was over the course of a game, season or career.

As Baseball Prospectus put it, “Pitchers who load the bases and escape are treated the same way as pitchers who strike out the side.” It’s for this reason that BP designed a new, all encompassing statistic for pitchers aimed at including as many explanatory variables as possible to explain a pitcher’s true performance level.

The new statistic is called DRA or Deserved Run Average and aims at quantifying just how much of the run scoring against a pitcher is directly attributable to their performance.

But it’s not like this is the first shot aimed at creating a pitcher’s super-stat. FIP (Fielding Independent Pitching) was invented and popularized as a statistic that used home runs, walks, strikeouts and hit-by pitch to measure a pitcher’s true output. It was used around the league as predictor for future success from players with substandard ERA numbers. For a time, it allowed the baseball industry to exploit an inefficiency in the market. DRA goes further.

DRA uses a run expectancy model similar to the one in The Book as linear weights to attribute the success of a pitcher through each event across the entire season. These events include basically anything that could occur during a given at-bat: walks, singles, doubles, home runs, strikeouts etc. It associates the given run expectancy with what the pitcher is doing at the time.

This means pitchers who strike out the side and those who walk three while doing it, are no longer treated the same. Pitchers who leave runners on base when leaving the game are charged with those errors and those entering a game with runners on are awarded for leaving them on base.

One of the most significant additions this statistic has to the baseball world is its ability to incorporate numerous variables with serious explanatory potential. Park effects, quality of defence, count in the at-bat, handedness of the batter and even weather are all things that DRA quantifies while measuring a pitcher’s success.

Another benefit is that DRA doesn’t distinguish between an earned run and unearned run; effectively eliminating scorer’s bias from the equation.

One limitation to this however is that DRA is not currently able to scoring across different eras. Thus, a 2000 Pedro Martinez DRA of 3.00 is not directly equivalent to a Clayton Kershaw DRA of 3.00. But the stat is hardly weeks old so there is still plenty of time to work the kinks out.

Being a product of Baseball Prospectus, this data is publicly available for anyone willing to go out and look for it. It’s conceivable teams have had this proprietary information and already knew everything this statistic is looking at.

But without a doubt, this advancement should be seen as a step forward for the sabremetric world at a time long after its moneyball revolution. While the metric may have some growing pains and problems to work out, it’s a bright day for the growing knowledge of baseball’s most important position.

Load Comments