[IrishTim]

Say you are using a radius of 2 (to remedy the sample size problem) from the closing spread in determining push probabilities. And assume we're dealing with a sport that doesn't have ties. Well then what do you do when you get to the 2, 1 and pick?

[Dark Horse]

Are you asking how to use push frequencies?

[IrishTim]

Sorry for being unclear. I'm asking how to calculate the push rate for the numbers near pk. Stanford Wong says that the best way to determine the historical push frequencies is to see how many games +/- 1 or 2 from the spread landed on a given number to broaden the sample size.

So if I'm looking for the push rate of the 5, I'd take all the games with a spread of 3, 4, 5, 6, or 7 and divide the number of games with those spreads where the favorite won by exactly 5 and divide it by the total number of games with those spreads.

My question was how do you deal with the number around pick, because obviously the "pk" can't push in basketball.

But now that I actually think about it for more than 20 seconds, I realize my question doesn't make any sense. Whether or not a game can push on the "pk" is irrelevant to the push rate of the 1 or 2 - all that matters is how often a game with a spread of pk pushes the 1 or 2. I was worried I had been systematically miscalculating the push rates for the 1 and 2 in all sports.

Thanks DH.

[skrtelfan]

Just ignore the 0 entirely and consider -1 to +1 as a half point move. So if you're using 2 points each way to broaden the sample size, or 9 half points total, for +1.5 you'd look at -2 to +3.5.

[LT Profits]

Just ignore the 0 entirely and consider -1 to +1 as a half point move. So if you're using 2 points each way to broaden the sample size, or 9 half points total, for +1.5 you'd look at -2 to +3.5.

^^^ This sounds like the best approach.