Dissecting Bullpen Usage

Bullpens are tricky creatures. They're awfully fickle, largely because of the extremely small sample sizes involved, and the self-selection of matchups created by managers. Furthermore, "traditional" stats---even the more advanced ones like K/BB ratio or basic ERA---often don't measure a reliever's true effectiveness. Consider the following scenario:

The Indians lead 3-2 in the 8th inning when CC Sabathia runs out of gas. Eric Wedge leaves him in two batters too many, and Sabathia gives up consecutive line-drive singles to begin the 8th. The opposing team has runners at first and third base, with no outs. Wedge brings in Reliever A.

Reliever A proceeds to give up a fly-out, then gets a ground-ball double play. However, the fly-out just deep enough for the runner at third base to score. The pitcher's raw line looks very good: 1 inning, o runs. Did he do his job?

Yes and no. His ERA will go down. Someone will cite that he has the ability to generate double-play ground-balls. And yet, the game is now tied. It doesn't matter who the tying run is charged to, the thing that matters is that it scored at all.

Thus, bullpens, while certainly fickle, aren't even able to be evaluated easily. Baseball Prospectus has some some excellent advanced statistics which more properly evaluate a reliever's contribution. However, there does not appear to be an easy way to measure or predict the advanced-statistic using traditional stats of any sort. Beyond the pitchers that are obviously very good, there are some that contributed more or less to their team than their raw statistics would suggest.

So how do you maximize your bullpen's effectiveness? What kind of pitcher do you want to bring in the game in the 8th inning of the above example? Intuitively, it makes sense to me that you want a reliever who's going to strike a lot of batters out. While many teams see the value in these types of relievers, I believe they are vastly misused. Here's how:

Using the above example, you want the reliever who you bring into the game in the 8th inning to be able to strike a lot of batters out. You don't just want any kind of out in that situation, you want a strikeout; any other kind of out risks tying the game in the form of a sacrifice fly or the like. Only with a strikeout can you guarantee that the runner from third does not score. Furthermore, because the score is close, the situation is very important, even though it's not the 9th inning. This is the ideal situation where teams should use their "closers" (who tend to be their best reliever, and probably the one most likely to get a strikeout).

Most teams bring in their closers to work the 9th inning. Usually, they bring in the closer when no one is on base, at the beginning of the inning. The closer needs to get three outs, but it does not matter how he gets these outs---in this situation, a strikeout is no more valuable than a ground-out or a fly-out. Thus, it's a waste to use your best strikeout-pitcher in this situation. Furthermore, wouldn't you want your best pitcher in the game when the game is most likely to be won or lost, such as in the situation above? If you choose to "save" your closer for only save situations, you are likely costing yourself some games which are lost prior to becoming save situations in the first place.

The best bullpens are the bullpens that teams stumble on to accidentally. The 2006 Tigers are a perfect example of this. Their closer, Todd Jones, was arguably their third or fourth best reliever, and he did not record many strikeouts. However, he did record outs anyway, and thus earned saves. However, their best relievers---and the relievers most likely to record a strikeout---were the set-up men, led by Joel Zumaya. Zumaya and company oftened preserved precarious leads in the 7th and 8th innings, thus leading to more save opportunities for Jones. Had their roles been reversed, Jones and Zumaya could have had the exact same amount of "personal success," but it would have translated to fewer victories for the Tigers as a team.

My hope is that something similar occurs with the 2007 Indians. I am very high on Fernando Cabrera. Witness his track record:

In his first full season as a reliever, in 2004 with AAA Buffalo, Cabrera pitched 75 innings, posting a 3.38 ERA, while striking out 92 and walking 42 (and giving up only 57 hits). The strikeout and hit rates were very encouraging, the ERA was solid (although remember how this can be misleading) and he walked too many. He also gave up 9 homers.

In 2005, also with Buffalo, Cabrera continued his dominance, and improved his control. He pitched 51 innings, allowing only 36 hits, a mere 3 homers, and struck out 68 while walking only 11. His ERA was 1.23. That's dominance.

With the Indians in 2005, Cabrera struck out 29 in 30 innings and allowed only one homer, posting a 1.47 ERA.

However, in 2006 Cabrera faltered. Despite his struggles---5.19 ERA, 12 homers allowed in only 60 innings---Cabrera's stuff remained in tact, as evidenced by his 72 strikeouts. He walked 32 and gave up 53 hits.

Much of Cabrera's problems were at the beginning of the season, possibly stemming from his participation in the World Baseball Classic.

Cabrera has the stuff and track record to be a very dominant reliever, striking out well more than one batter per inning and giving up few hits. He walks too many, but has shown the ability to improve on that. Furthermore, he's only 24 years old.

Cabrera is the ideal candidate to bring in the above 8th-inning scenario. If the Indians are lucky, Cabrera could be the key cog in a much improved bullpen.