Sunday, August 12, 2007

The Bullpen

Especially of late, the 2007 season has been a slight disappointment after the magical 2006 affair. While the offense has been at times dominant, the pitching has taken a turn for the worse. Given that the pitching is considered the young core of the team, this has been a tremendous disappointment in a lot of ways.

The bullpen, in particular, has taken a big step back. In order to take a look at this, I'm going to look at some of the Win Expectancy numbers from the guys at Baseball Prospectus.

Before going into this, I want to try my best to explain the concepts of "above average" and "above replacement. We'll take a look at four players and calculate their hits above average and hits above replacement.

Player A 15 PA, 5 hits
Player B 18 PA, 4 hits
Player C 11 PA, 7 hits
Player D 20 PA, 1 hit

To calculate what is average, we should calculate the average amount of hits per plate appearance for all of the players.

Player A 0.3333 H/PA
Player B 0.2222 H/PA
Player C 0.6364 H/PA
Player D 0.0500 H/PA
Average 0.2656 H/PA
Rep Level 0.2000 H/PA

Replacement level is defined as a player that can be found on the waiver wire. I really just plucked that number out of nowhere to go through these calculations. Now we want to find out how many hits/plate appearance above average and multiply by plate appearances to find hits above average and replacement.

Player A 0.0677 H/PA above Average (1.016 HAA)
0.1333 H/PA above Replacement (2.000 HAR)
Player B -0.0434 H/PA above Average (-0.782 HAA)
0.0222 H/PA above Replacement (0.400 HAR)
Player C 0.3608 H/PA above Average (3.969 HAA)
0.4364 H/PA above Replacement (4.800 HAR)
Player D -0.2156 H/PA above Average (-4.312 HAA)
-0.1500 H/PA above Replacement(-3.000 HAR)

I really wanted to go through that, because I've always found that people shy away from numbers like this and to maybe show some trends. Average players with lots of playing time can rack up numbers over replacement level, while good player without much playing time will have a better time getting above average compared to replacement.

Now that we have that out of the way, I wanted to take a peak at the Win Expectancy of our bullpen. Win expectancy was created by Clay Davenport at Baseball Prospectus, and essentially takes the guessing out of inherited runner numbers. For a full explanation, go here. The basic concept is that he used historical data for situations and how often teams win based on the score and how many runners are on base, and made adjustments for team strength. The totals are compared to average and replacement level, and given an adjustment for who the players face in the lineup (For example, a player would get more credit for retiring Sheffield/Magglio/Guillen than one who takes care of Casey/Monroe/Inge). Players who come into mop up situations will likely linger around zero, while pitchers who close out the game or come into close situations

Oh yeah, we were going to look at the Tigers bullpen. I guess we should do that. Here are the 2007 lineup adjusted numbers for Win Expectation above an average reliever.

Bobby Seay 0.995
Todd Jones 0.947
Joel Zumaya 0.620
Chad Durbin 0.446
Tim Byrdak 0.093
Aquilino Lopez -0.019
Eulogio Delacruz -0.031
Jose Capellan -0.064
Yorman Bazardo -0.066
Macay McBride -0.111
Zach Miner -0.151
Wil Ledezma -0.322
Jose Mesa -0.503
Jason Grilli -0.513
Fernando Rodney -0.539

What does this tell us? Todd Jones has done well closing out games. He gives up runs in bunches, which means that the last couple runs he gives up in an appearance probably don't decrease the team's shot at losing much. Bobby Seay has been used properly this year, facing primarily lefties. We need Zumaya back. I wouldn't think Durbin would rate this high. I'm going to look into seeing if his starting innings are skewing the analysis. The bullpen overall has been bad. Jason Grilli has been as bad as everybody this side of Jim Leyland thinks, and the sub-100% Fernando Rodney was terrible. Luckily, he seems to be better now.

For comparison, here's a look at our 2006 numbers.

Joel Zumaya 3.978
Fernando Rodney 1.149
Todd Jones 0.791
Jamie Walker 0.371
Roman Colon 0.095
Jordan Tata 0.079
Chad Durbin 0.037
Bobby Seay -0.002
Zach Miner -0.060
Colby Lewis -0.068
Wil Ledezma -0.075
Chris Spurling -0.102
Andrew Miller -0.116
Mike Maroth -0.150
Kenny Rogers -0.288
Jason Grilli -0.393

Look who was our worst reliever last year... As with Durbin this year, I'm not sure Maroth and Rogers are rated fairly. If I remember correctly, Rogers was used in the last game of the season out of desperation and didn't pitch too well. One could make a case that Zumaya's year last year was the best for a Tigers reliever in the franchise's history. Considering he'd be going up against 1984 Willie, that's saying something. Overall, the three guys counted on to keep leads did the job. Ledezma and Grilli on the surface pitched better, but clearly struggled with inherited runners and possibly benefited a bit from Zumaya cleaning up their mess.

Overall, the bullpen has been much worse this year. As shown above, the notion that Jason Grilli has been overused is probably correct.

I've had Internet issues of late, and will try to get those straightened out so I can update this more often. Soon, I'll take a further look at another tool Davenport has in these rankings and go through and try to get to the bottom of the options/service time mystery.

No comments: