D-Train (barely) averts Disaster Start


Every time I watch Dontrelle Willis pitch, I think of Fred Thompson's line in "The Hunt for Red October" right after the (ridiculously obvious) stock footage of the botched carrier landing: "This business will get out of control. It will get out of control and we'll be lucky to live through it."

Early in today's Tigers-Royals game, I Tweeted this:

    Uh-oh ... I think D-Train's heading for a Disaster Start (definition to follow) ...

Basically, I did that so I'd be forced to come up with a new junk stat (as Bill James calls something like this) ... What is a disaster start? What would it look like in the box score?

The most simple and obvious idea: Runs > Innings Pitched.

I love simple, but that's too simple. While the system for awarding wins to pitchers is silly, it's not going to change and I don't want to see a pitcher getting hung with a disaster start and getting credit for a win. It wouldn't happen often, but it would happen: five innings and six runs, whatever.

I suppose we could make whatever rules we like, with the caveat that you can't get a disaster start and a win in the same game, but I don't like the rules for a win and I don't want to tie my brilliant invention to those silly rules. Also, in some sense when a pitcher goes five innings he's done something good, right? Saved the bullpen, just a little bit? Sometimes a manager leaves a pitcher in without his best stuff, just to eat an inning or two. Or maybe the manager's just an idiot. Either way, it's not the pitcher's fault.

So I'm happy with this definition: Less than 5 innings, more runs than innings.

By that definition, Willis has

  • 3 disaster starts (out of 7) in 2008,

  • 2 disasters (of 7) in 2009, and

  • 0 disasters (of 2) in 2010.

That's 5 disasters in 16 starts, which really doesn't seem so bad. Maybe we do need to be tougher?

I should have thought of this first, but we could tie disasters to Bill James' Game Score. The downside is that I'm not smart enough to figure Game Scores in my head. The upside is that we publish them in our box scores and Baseball-Reference.com has them for almost every start going back 50 years.

Another nice thing: The scales are intuitive. A brilliant start (with lots of strikeouts) will top out around 100. An average start is around 50. And if we define disaster start as 25 or less, we wind up with a satisfactory list.

It's not the same list, though. Here are three of Willis' gems last year:

5.0 innings, 10 hits, 7 runs, 2 walks, 1 strikeout

2.1 innings, 0 hits, 5 runs, 5 walks, 3 strikeouts

3.2 innings, 6 hits, 6 runs, 6 walks, 1 strikeout

According to my simple method, the second and third of those were disasters, but the first wasn't because he did manage to last five innings. But the Game Scores for those three are 18, 35, and 18.

I suppose the second start gets a 35 because he didn't give up any hits, but to me that's a disastrous start. I'm a big fan of Game Scores. But like a lot of portmanteau stats, this one might not work real well at the margins. And getting knocked out in the third inning after giving up five walks and five runs is pretty marginal.

So I'm sticking with simple. And I'm sticking with my prediction: Willis is going to lose his job this spring, after a couple of disaster starts. The Tigers are finished with the Royals for a while. Now they get to face the Angels and the Rangers and the Twins. Watch those box scores. Disasters waiting to happen.