May 17, 2007
Double Header Strategy
Bill James many years ago wrote about Earl Weaver employing a strategy of using his "B" lineup in the first game of a double header, and his "A" team in the second game. Since most teams do the opposite, Weaver thought he'd gain an advantage with the strategy. If he got lucky and won the first game, the team was more likely to sweep. If he lost the first game, he was more likely to split. I'm not sure how the probabilities work out; I assume it's very dependent on the strength of the opposing team.
Terry Francona appears to be using this strategy today. In game one, Hinske, Mirabelli and Cora are all starting, as is Tavarez. Leyland is using his regular lineup. We'll see how this works out over the day.
Posted by David Pinto at
12:29 PM
|
Strategy
|
TrackBack (0)
Although he's due for a major regression, I would argue that Cora is an upgrade over Pedroia at this point in the season.
Intuitively it seems to me that the better team should want to match strength against strength, the worse team should prefer Weaver's approach, trying to match his strength against the opponents' weakness.
I checked this on the assumption that the better team scores on average 6 runs/game with its better squad, 5 with its weaker squad, while the inferior team scores 5 runs/game with its better squad and 4 with its weaker. Obviously these are simplistic. Anyway, I used the Pythagorean method to find expected wins. This confirms that the superior team gets more expected wins when the stronger squad is matched with the stronger. But the difference is very small (the better team gets 1.20 expected wins instead of 1.19).
Something similar I've thought about. If we were to take a theoretical division each with equally-matched #1 pitchers, #2 pitchers, and so on down the line to a #5 guy, the ideal way to set up the rotation, assuming the other teams go with the standard 1-5 starters, might be to start your #5 guy first against your opponent's top starter, then go with #1 against his #2, #2 against his #3, etc. That way you've got an advantage in 4/5 of the games. Granted, this is a simplistic model, but it's worth thinking about.
If there are A and B teams, the Mets seem to be using their C team today. Julio Franco's playing third!
Hi Adam - I looked at this a couple years ago - check it out:
http://www.hiremetheo.com/wordpress/?p=68
The table and discussion of your question is a little ways down in the post. Bottom line, is that staggering your rotation like you suggest would appear to add .13 extra wins every 5 game set, in a league with teams that have starters with ERAs of 2, 3, 4, 5, and 6.
Earl Weaver's explanation sounds like crap since the reverse would also be true, though not on chronological order... if the other team got lucky winning the second game, they're more likely to sweep.
But I agree that it might sometimes be the right choice. Though I'm not sure if it's necessarily the better team that wants AvsA and BvsB, it might be the deeper team. Their B team would still be likely to lose to the opponent's A, but in BvsB, they'd have an edge.
