from Amazon
Friday, April 27, 2007
When to bring in a reliever
There’s alot of good info in The Book on this. Here are a few more considerations that researchers can try to analyze:
(Note: cross-posted to SOSH)
A 3 run lead in the bottom of the 9th is a slightly higher leverage than one in the top of the 9th. The reason is that if the Redsox are at home, they still get to bat in the bottom of the 9th.
I wouldn’t term either situation “low leverage”. The one Papelbon came in had an LI of 1.02:
http://www.fangraphs.com/plays.aspx?date=2007-04-26&team=Orioles&dh=0&season=2007
That’s the very definition of average. He helped himself by getting an out right away which did in fact turn the game at that point into a fairly low-leverage situation (0.57).
Regardless, if you haven’t pitched in a while, the threshhold at which you should enter the game should decrease. You could create a basic model of the form:
7th inning, LI greater than 3.0
8th inning, LI greater than 2.5
9th inning, LI greater than 1.5
And then, for every day off he has, reduce the above numbers. So that maybe if you have 4 days of rest, the above numbers become 2.5, 2.0, 0.75, or some such. (All numbers for illustration only.)
Beyond this optimization that you’d need to perform (different for each team, based on the rest of the bullpen), you’d also want to factor in the warming-up time. A reliever would need at least a 2-batter lead time, which could cause the LI to swing wildly, especially if it’s either the 9th inning, or there’s 2 outs in an inning.
In short, once you have the numbers:
http://www.insidethebook.com/li.shtml#18
You construct a model of usage, based on human conditioning.