Smyly Among Top AL Relievers, But Is He Being Used Optimally?

In a recent post, I used the RE24 statistic to measure batting contribution including situational hitting.  The statistic is appealing because it gives batters more credit for hits with runners on base than for hits with the bases empty.  This concept can also be applied to pitchers, relievers in particular.

Statistical evaluation of relievers is difficult for a couple of reasons:

1. They pitch so few innings that their statistics can be influenced heavily by a couple of really bad outings.
2. Their actual value depends on game situations more than any other player (this problem will be addressed here)

Using ERA to evaluate relievers is problematic because relievers often make appearances with runners on base and give up other pitcher's runs. So a pitcher could have a low ERA without actually being that effective. FIP ERA which is based on walks, strikeouts and home runs allowed rather than runs allowed is better but it still does not consider the game environments in which a reliever pitched.

The RE24 metric estimates the number of runs a pitcher saved or cost his team based on his numbers of singles, doubles and all other events allowed including outs.  It also considers the situations in which these events happened.  For example, if Tigers southpaw Phil Coke enters a game with two men on base and nobody out and retires the side he will get more credit than if he comes in with the bases empty.  Coke gets more points in the first scenario because there was a greater potential for run scoring.  Thus, Coke saves the Tigers more runs if he frequently pitches well with runners on base than if he always starts an appearance with the bases empty.

The American League RE24 leaders among relievers are shown in Table 1 below.  Texas Rangers right hander Tanner Scheppers leads the league with a RE24 of 12.9.  This means that he has saved the Athletics an estimated 13 runs compared to an average pitcher with the same number of outs. This is not too surprising given his 0.64 ERA and relatively large workload (28 innings).  There are two Tigers in the top ten - Drew Smyly (10.6) and Joaquin Benoit (8.4).

Table 1: AL RE24 Leaders Among Relievers

Pitcher	Team	RE24
Scheppers	Tex	12.9
Smyly	Det	10.6
Cecil	Tor	9.7
Ross	Tex	8.7
Claiborne	NY	8.4
Hunter	Bal	8.4
Benoit	Det	8.4
Tazawa	Bos	8.1
Uehara	Bos	7.7
Coello	LA	6.8

  Data source: FanGraphs.com

Other Tigers relievers with 10 or more innings are listed in Table 2 below.  Darin Downs, Jose Valverde and Jose Ortega have all saved the team runs while Phil Coke and Al Alburquerque have been below average. This might explain why Alburquerque was briefly demoted to the minors.
  

Table 2: RE24 for Tigers Relievers

Pitcher	RE24
Smyly	10.6
Benoit	8.4
Downs	3.5
Valverde	2.4
Ortega	0.9
Coke	-1.4
Alburquerque	-1.7

Data source: FanGraphs.com

 

Smyly has been the team's most effective reliever, but has he been used in situations which help the team win?  Table 3 lists the average Game Leverage Index (GmLI) of the Tigers relievers when they enter a game.  Leverage Index, a concept developed by Tom Tango, measures how critical a given plate appearance is in determining the final result of the game by looking at the difference in win probability between the best and worst case scenarios.

For example, suppose Smyly enters a game in which the Tigers have a nine-run lead with nobody on base in the ninth inning.  In that situation, there is little difference between giving up a home run and getting the batter out on the probability of winning the game.  This is an example of a low-leverage situation. However, if Smyly comes into a game with two on and nobody out and a one-run lead in the ninth inning, then allowing a home run will be much more damaging than getting the batter out.  This is a high-leverage situation.

A value of one is assigned to an average game situation.  Higher-leverage scenarios have values of more than one and lower-leverage scenarios have values less than one.  These values are averaged together to get a pitcher's index for the season.  Coke has a GmLI of 2 which means that, on average, he has entered games in situations with twice the impact of an average at bat.  Smyly, on the other hand, has been used in only average situations (1.0 GmLI).  This is low for a reliever, especially a good one, as they are typically used at important moments in games.

Table 3: Leverage Index for Tigers Relievers

Pitcher	GmLI
Coke	2.0
Valverde	1.7
Benoit	1.4
Downs	1.3
Alburquerque	1.2
Ortega	1.0
Smyly	1.0

Data source: FanGraphs.com

This is fuel for the fire for those who think manager Jim Leyland is not using Smyly optimally.  Smyly is saving the team runs, but it's not usually happening at the most crucial points of games.  There is probably more to the story as I suspect it is an organizational decision to stretch out Smyly as a long reliever, so he can be ready to start if necessary.  One can certainly question though whether he would have more value to the team as a back-end reliever than as a starter in waiting.