NBA: Weighting games or not for line projections?

[tool21]

I was thinking about throwing some weights into my line making program but really don't know if it's worth it. At first I was all about it but then after thinking more more about it i became hesitant.


Take for example Boston game 1 against Atlanta. If i had weighted according to most recent games, Boston's numbers would be wrong because the last 5 or so games they really didn't play their starters. Those games in which their starters didn't play i would be weighting the most and probably bet on some false lines. My question is is it worth weighting these stats or would it just be false information? The more I think about it the more i lean towards using just averages although using weights sounded like a great thing to do at the time.

[Ganchrow]

I was thinking about throwing some weights into my line making program but really don't know if it's worth it. At first I was all about it but then after thinking more more about it i became hesitant.


Take for example Boston game 1 against Atlanta. If i had weighted according to most recent games, Boston's numbers would be wrong because the last 5 or so games they really didn't play their starters. Those games in which their starters didn't play i would be weighting the most and probably bet on some false lines. My question is is it worth weighting these stats or would it just be false information? The more I think about it the more i lean towards using just averages although using weights sounded like a great thing to do at the time.

I think you've answered your own question.

If you're going to weight games (perhaps using an exponential moving average) you'd first need to make sure that the games deserved to be equally weighted after netting out the time effect. Clearly, in the example you've given, this would not have been the case as there'd be transient factor (proximity to playoffs) that would have exerted a considerable effect on game outcome.