I'm a baseball fan living in New York City. In between long tirades about the New York Yankees and the national pastime in general, I'm a graphic designer.
Sometimes it appears that managers have taken the information revealed by the sabermetric revolution to heart. Falling pitch counts
are one area, the increased emphasis in On Base Percentage is another.
But there's one area in particular where the statistical evidence points towards exactly the opposite of what most managers seem to do. And it's something that often costs them the ballgame. I'm talking about the tendency of a manager to let a tiring pitcher start off an inning in which he will be pulled if he gets into trouble.
First off there's the pitch count issue; the current working theory is that pitches beyond a certain threshold (say, 120 pitches) are more likely to lead to ineffectiveness and injury. That's a topic for another day, however.
In Bill James's 1987 Baseball Abstract, James reported on research done by Gary Skoog with regards to how many runs a team would score based on a certain situation (i.e., runner on first, one out). Here is that matrix:
Expected number of runs 0 outs 1 out 2 outs
no one on 0.454 0.249 0.095
runner on first 0.783 0.478 0.209
runner on second 1.068 0.699 0.348
runner on third 1.277 0.897 0.382
runners on first & second 1.380 0.888 0.457
runners on first & third 1.639 1.088 0.494
runners on second & third 1.946 1.371 0.661
bases loaded 2.254 1.546 0.798
What these numbers mean is that at the start of 1000 innings (0 on, 0 out), teams can be expected to score 454 runs. With a runner on first and no outs, that expectation rises to 783 runs per 1000 innings; with one out and nobody on, that expectation falls to 249 runs per 1000 innings.
The difference between those two states (runner on first, 0 out and no runner, 1 out) is greater than the expectation of runs at the start of the inning. The leadoff batter is THE MOST IMPORTANT BATTER. When he gets on, teams score runs--at LEAST three times as many runs as if the first batter makes an out.
So why in the HELL would a manager risk letting his (tiring) pitcher screw up the start of an inning? I'm sure it has something to do with instilling confidence in your starters (blah, blah, blah), wanting to save your bullpen (blah, blah, blah), or some bullshit along those lines. But unless your only relief options have been soaking in kerosene or are named Bobby Ayala
, this tendency strikes me as an incredibly backward way of thinking.
I realize the limitations of this matrix. Since these are averages, certain situations, such as having Rey Ordoñez leading off your inning (lower run expectancy at 0,0) or Barry Bonds (much, much higher), might dictate a change in strategy. Managers might also choose their pitcher here based on the platoon advantage (righty pitching to righty), but that advantage is only on the order of 20-25 points of batting average. And then there's the disadvantage of bringing in a reliever with men on base: pitching from the stretch, limited pitch selection, wild pitches... so much more can go wrong with a man on base.
I didn't watch Yankees game today, but Ted Lilly's pitch count was nearing 120 as he went back out for the 8th inning. He had a 4-1 lead, and had pitched a good ballgame, striking out 9 batters and allowing only 5 hits. Joe Torre had any number of options to get two innings out of the bullpen, including one of the best closers in the biz, Mariano Rivera. Rivera and setup man Mike Stanton have been overworked, but even if Torre didn't want to use them, he should have put his pen in the best situation possible to finish the game. And that means starting with a clean slate at the beginning of an inning.
Lilly allowed a double to Steve Cox and was done for the day. With the number of pitches he'd thrown, he wasn't going to throw a complete game, so why the hell was he still out there? Brian Boehringer came in, allowed a double and a walk before yielding to Stanton, who, with the roof on fire added just a little more Ronsonol, and when the ashes settled the Yanks were down 5-4.
It's easy to sit here and second-guess an individual set of decisions a manager makes. That's not my intention; Joe Torre has won four more World Championships than I have, and he's a fine manager. But he, and dozens of other managers I've watched since I first read Bill James, keep making this mistake, and it drives me crazy.
[Somebody whose site I discovered ('Rhoids Baseball, whatever the heck that is) in searching for the above table has updated the data based on more current run scoring tendencies. Here is the revised table
. The current values of 0 on, 0 out (0.58), runner on 1st, 0 out (0.98), and 1 out, 0 on (0.31) are all higher, but their ratios are similar enough that they don't change my argument.]