Do Managers Properly Use Their Relievers?
The question of how well managers use their relievers is one I have been asking myself for a while and a question I finally tackled. I alluded to my perception of a manager misusing his bullpen when Joe Torre sent out Kyle Farnsworth in the 8th inning to face the heart of the Boston Red Sox lineup and let Mariano Rivera, his best reliever, face the bottom of the Red Sox lineup in the 9th inning because he had to get Rivera another save.
Since I wanted to know if managers really put their best relievers against the best hitters, I did a little research. I decided to not only look at if the best relievers were pitching against the best hitters, but to also include if managers put their relievers in the highest leverage situations. The statistics I used to determine this were WXRL (Win Expectation Added Over Replacement Level), Leverage, and the collective faced batters' wOBA. wOBA is weight on-base average and is a statistic developed by Tom Tango, Michael Lichtman, and Andrew Dolphin and is a more accurate statistic than OPS because it does not treat OBP and SLG equally. For the purposes of my research, I used the approximate formula of OBP*2+SLG/3 to get wOBA.
Then I took five relievers from each of the thirty MLB teams. To choose the relievers, I used their saves, holds, and as a tiebreaker innings pitched. Relievers who had also started were excluded from the data. Hopefully, the one hundred fifty pitchers were enough to overcome any sampling errors, but during the All-Star Break I will go back and look at all the relievers in the majors, but for right now, I think the one hundred fifty relievers will do.
Getting a handle on how good managers use their bullpens required me to do a correlation between WXRL and Leverage and also WXRL and wOBA. The higher the positive correlation, the more the two variables are linked in the same direction. In simpler terms, a high positive correlation would tell us that the managers put their best relievers in the highest leverage situations and also put their best relievers against the opponents' best hitters.
The results are sobering indeed. For the one hundred fifty pitchers studied, the correlation between their WXRL and their Leverage scores is .499. The correlation between WXRL and wOBA is -.052, which is basically zero which is basically no correlation at all.
There are a few explanations for the low correlation between WXRL and Leverage. Bad luck could be the culprit or as I mentioned earlier, there could be a sampling error which we will find out when I look at all the relievers in the major leagues. Another explanation could be it takes half of a season for a manager to realize what he has in his bullpen and what his pitchers are capable of. If that is so, then it does not speak highly of managers.
Another explanation, and the one I am most likely to believe, is that managers really have no idea how to incorporate a game's leverage into selecting which reliever to call from the bullpen or they are unwilling to make minor changes in the bullpen to keep in line with their pitchers' performances. Because a player is being paid to be an 8th inning set-up man does not mean he should be left in the role if he is struggling. More imagination is necessary from baseball managers in this respect.
As for wOBA, the correlation of -.052 makes it pretty obvious that managers do not care which batters are coming up in a particular half of the inning. Instead, their only focus is on which inning it is.
Still, there is a bright side. Although overall, MLB teams use their relievers horribly, there are nine teams with a correlation above .800 in terms of WXRL and Leverage. Leading the way with a .981 correlation are the San Diego Padres. The other eight teams are the Boston Red Sox, Minnesota Twins, Los Angeles Angels, Florida Marlins, New York Mets, Washington Nationals, Cincinnati Reds, and the Houston Astros. As you can see, this list has both good teams and bad teams on it so just because a team is doing well does not mean they deploy their bullpen properly.
If the overall low correlation between WXRL and Leverage and WXRL and wOBA holds up when I extend the sample to include all receivers, then it will indicate that looking at Leverage alone to tell how managers use their bullpen is not enough. Who cares how well a manager puts his closer in high leverage situations if his set-up man is the best reliever on the team?
Stay tuned for the updated list and correlations.
Since I wanted to know if managers really put their best relievers against the best hitters, I did a little research. I decided to not only look at if the best relievers were pitching against the best hitters, but to also include if managers put their relievers in the highest leverage situations. The statistics I used to determine this were WXRL (Win Expectation Added Over Replacement Level), Leverage, and the collective faced batters' wOBA. wOBA is weight on-base average and is a statistic developed by Tom Tango, Michael Lichtman, and Andrew Dolphin and is a more accurate statistic than OPS because it does not treat OBP and SLG equally. For the purposes of my research, I used the approximate formula of OBP*2+SLG/3 to get wOBA.
Then I took five relievers from each of the thirty MLB teams. To choose the relievers, I used their saves, holds, and as a tiebreaker innings pitched. Relievers who had also started were excluded from the data. Hopefully, the one hundred fifty pitchers were enough to overcome any sampling errors, but during the All-Star Break I will go back and look at all the relievers in the majors, but for right now, I think the one hundred fifty relievers will do.
Getting a handle on how good managers use their bullpens required me to do a correlation between WXRL and Leverage and also WXRL and wOBA. The higher the positive correlation, the more the two variables are linked in the same direction. In simpler terms, a high positive correlation would tell us that the managers put their best relievers in the highest leverage situations and also put their best relievers against the opponents' best hitters.
The results are sobering indeed. For the one hundred fifty pitchers studied, the correlation between their WXRL and their Leverage scores is .499. The correlation between WXRL and wOBA is -.052, which is basically zero which is basically no correlation at all.
There are a few explanations for the low correlation between WXRL and Leverage. Bad luck could be the culprit or as I mentioned earlier, there could be a sampling error which we will find out when I look at all the relievers in the major leagues. Another explanation could be it takes half of a season for a manager to realize what he has in his bullpen and what his pitchers are capable of. If that is so, then it does not speak highly of managers.
Another explanation, and the one I am most likely to believe, is that managers really have no idea how to incorporate a game's leverage into selecting which reliever to call from the bullpen or they are unwilling to make minor changes in the bullpen to keep in line with their pitchers' performances. Because a player is being paid to be an 8th inning set-up man does not mean he should be left in the role if he is struggling. More imagination is necessary from baseball managers in this respect.
As for wOBA, the correlation of -.052 makes it pretty obvious that managers do not care which batters are coming up in a particular half of the inning. Instead, their only focus is on which inning it is.
Still, there is a bright side. Although overall, MLB teams use their relievers horribly, there are nine teams with a correlation above .800 in terms of WXRL and Leverage. Leading the way with a .981 correlation are the San Diego Padres. The other eight teams are the Boston Red Sox, Minnesota Twins, Los Angeles Angels, Florida Marlins, New York Mets, Washington Nationals, Cincinnati Reds, and the Houston Astros. As you can see, this list has both good teams and bad teams on it so just because a team is doing well does not mean they deploy their bullpen properly.
If the overall low correlation between WXRL and Leverage and WXRL and wOBA holds up when I extend the sample to include all receivers, then it will indicate that looking at Leverage alone to tell how managers use their bullpen is not enough. Who cares how well a manager puts his closer in high leverage situations if his set-up man is the best reliever on the team?
Stay tuned for the updated list and correlations.
posted by David at
9:30 PM
0 Comments:
Post a Comment
<< Home