Comments on: Pitcher Win Values Explained: Part Three

By: brent in Korea

brent in Korea — Wed, 14 Jan 2009 02:00:04 +0000

At Battersbox.ca, there was a lot of support for having “the fifth starter” be taken over by two young players. The starter would only go four or five innings. The “next starter” would be brought in to go four or five innings. Does anyone have any thoughts on this?

By: azruavatar

azruavatar — Wed, 14 Jan 2009 01:48:08 +0000

Dave — The link doesn’t answer where the 38% really comes from. Control for the offense, defense & bullpen (as much as possible). Then. . . I’m a little fuzzy how we then figure out how often that team would win with a replacement level starter (since we shouldn’t yet know what a replacement level is). Are we basing this off a set of probabilities or empirical sampling? I’m not arguing that this number is wrong, I just looking for the link between the qualitative description and the quantitative result. I don’t understand the methodology used to derive it.

By: Dave Cameron

Dave Cameron — Tue, 13 Jan 2009 23:53:02 +0000

League average changes by year and league, but we want to keep replacement level fixed, so we use the percentages to calculate different replacement levels for different years/leagues.

By: CMC_Stags

CMC_Stags — Tue, 13 Jan 2009 23:21:11 +0000

I think the winning percentage is used to calculate the necessary ERA (or FIP for this purpose). If you know what the average starting pitchers FIP is and you assume both team’s defense and offense are the same, then you can use Pythag to determine a Replacement Level ERA/FIP/tRA etc.

By: azruavatar

azruavatar — Tue, 13 Jan 2009 22:28:12 +0000

Any reason to use the winning percentage and not the associated ERA (5.03 in the Tango thread found by smith) in the calculation?

By: Samg

Samg — Tue, 13 Jan 2009 22:11:44 +0000

And the “winning %” is referring to the fact that a team full of players of that level would win at that rate?