Say Im pricing a baseball prop that is HRE (hits plus runs plus errors) in a single game. I can come up with a number that is normalized for the opponent and ballpark. My problem is say I get something like 22.156, and the prop is 22.5 o/u -115. How can I tell if thats worth a bet? I normally use poisson, but in this (and plenty others) example, poisson doesn't fit. Ive been trying to teach myself bayes theorem, but Im not even sure that would be applicable here. Any ideas? Thanks in advance.
You could try a negative binomial which can handle over-dispersion. But the importance of fitting a distribution pales in comparison to park/weather correction; you can never put too much effort into those areas.
TomG- how's the fantasy market treating you these days? I dipped into that earlier this year, but found it was too big of a time investment for now.