Pricing NBA In-Game Middles

[bookie]

Does anybody have an idea, or know of a post, that suggests a rule of thumb way for determining push frequencies for in-progress games?

I think Ganchrow's half-point calculator is based on the idea of using only results that fall within a point or so of a given line. So the question is of all the games where the line was three what % fell X, Y, Z. Using this methodology you can price a result as having, in a rough rule of thumb way, about a three and a half percent chance of falling.

Does that rule of thumb hold when you're buying at different points in the game? So, for example, I have SA +3.5 and Oklahoma +3 tonight (it's half time of that game as I write with OKC up 8 and the HT line -5.5). Anyway, using the rule of thumb my five middle numbers and side number should make my chances to win about 16%, but that doesn't seem right to me...seems a little high.

Any helpful comments on how to approach this kind of evaluation?

[MonkeyF0cker]

Full game push frequencies are virtually irrelevant once the game begins.

You need to ask yourself how you'd figure out what the fair line is when Team A is up by X points with Y time remaining against Team B (among other factors). Then you could extrapolate the fair price of an alternate spread from a resultant scoring distribution.

[mathdotcom]

Monkey wearing his professor's cap tonight

[wrongturn]

Does anybody have an idea, or know of a post, that suggests a rule of thumb way for determining push frequencies for in-progress games?

I think Ganchrow's half-point calculator is based on the idea of using only results that fall within a point or so of a given line. So the question is of all the games where the line was three what % fell X, Y, Z. Using this methodology you can price a result as having, in a rough rule of thumb way, about a three and a half percent chance of falling.

Does that rule of thumb hold when you're buying at different points in the game? So, for example, I have SA +3.5 and Oklahoma +3 tonight (it's half time of that game as I write with OKC up 8 and the HT line -5.5). Anyway, using the rule of thumb my five middle numbers and side number should make my chances to win about 16%, but that doesn't seem right to me...seems a little high.

Any helpful comments on how to approach this kind of evaluation?

Definitely, the push possibility of points near current market line increases as game progresses. I am not sure exactly how much, but I simply use 10% in 4Q.

[bookie]

Thanks for comments WrongTurn and MonkeyF0cker...

Does anybody know of any in-game line archives?

[MonkeyF0cker]

You can get in-game line archives (from line services like DB). It's probably not going to help you though since you're not going to know for sure what point in the game they originated. You generally have to use other means to test the viability of an in-game model.