A’s Lineup Analysis
It hardly seems like a year has passed since I first wondered about optimizing lineups after reading an interesting article by Cyril Morong at Beyond the Boxscore on the value of spots in the lineup.
I couldn’t have predicted the avalanche that I’d help create, after my work was noticed by Ken Arneson, and then David Pinto created a handy tool for automating it. You just never know when an idea will take off, I guess. I’m hoping that my forays into umpire statistics will also create some discussion of an often ignored aspect of the game, but that’s neither here nor there.
To get to my point, it’s now that time of year when boredom from the lack of games drives some of us to ponder things like, oh, how many more runs would the A’s score over a season if they batted Bobby Crosby 8th instead of 7th?
I got down to business and started running lineups through Pinto’s tool, and here’s what I found.
—–
First off, let’s establish the accuracy of the lineup tool.
Last season, the A’s scored 771 runs, or 4.759 per game. A common lineup they used was the following:
1. Kendall
2. Kotsay
3. Bradley
4. Thomas
5. Chavez
6. Payton
7. Swisher
8. Crosby
9. Ellis
That lineup was expected to score 5.021 runs per game, so they underperformed the model by .262 runs. The best lineup that could have been made from those players would have scored 5.120 runs, and the worst would have scored 4.814. Obviously it’s just overestimating the runs scored by a little. I’ll try to allow for that in the next step.
Now that we have a general idea of how close the tool comes to estimating runs scored based on the lineup, let’s brace ourselves, and feed it a projected 2007 lineup that’s missing Thomas and Payton… I used numbers from the Grand Unified Projection.
1. Kendall
2. Kotsay
3. Bradley
4. Chavez
5. Swisher
6. Piazza
7. Johnson
8. Crosby
9. Ellis
That’s expected to score 5.130 runs per game, not bad at all. Tops is 5.186, low is 5.014. That’s better than what last year’s lineup was expected to do, in fact. If we account for the overestimation, I’d feel safe saying that this lineup could score perhaps an average of 4.87 runs per game, or 789 runs on the season.
Using the runs allowed estimate that devo of AN came up with, 711, that would put the A’s at a pythagorean record of… hang on let me get my calculator.
89-73 for 2007
That’s really about how many wins I’d expect from this team. It’s not great, but it will contend for a playoff spot. If Rich Harden actually stays healthy, though, I’d expect 3-5 more wins, perhaps.
If, for some reason Oakland performs up to the average level the model expects, 5.100 runs per game, they’d be expected to score 826 runs and go 92-70. Without making an adjustment for apparent overestimation, the top record would be 831 runs, for 93-69, and the low would be 812, for 91-71.
It sounds like it’s safe to say that the A’s should reasonably win 88-92 games next year, by this method.
Much of this depends on the pitching staff, though. Last season they allowed 727 runs, so I’d have to think a 16 run improvement, with losing Zito, would have to come from Rich Harden. If they can’t make up the difference with Blanton and Haren improving, and getting a healthy Harden, and gave up perhaps 745 runs, their expected record would drop to 85-77, which almost certainly won’t be good enough to contend in the AL West this season.