[rfr3sh]

So far with my model, my totals and spreads are very similar to the books lines for the most part, however I have noticed with games with high totals (over 210), they are usually way off, for example the goldenstate game tonight total is 214.5 right now at pinny, however my model only produces a total of 206, I have no clue why there is such a difference and it is always with the high total games. I have been working on this for the past 3 days and everything I tried hasn't worked. I know compared to the league averages, spurs are significantly above average offensively right now and goldenstate is significantly worse defensively, and my numbers reflect that, however not enough I guess

anything specific I should look over again?

[Pokerjoe]

I used to be in your shoes, and it took a while to figure it out, which means my silence (and everyone else's) will be loud. Nothing personal, honestly. Besides, the fun's in the figuring. Good luck with it.

And I'm not meaning to be a smartass in this post, btw. Just explaining why I don't think anyone will help, so you don't sit here thinking no one likes you or something, LOL.

[rfr3sh]

joe I understand, I actually enjoy the challenge, more then betting
I know that with the effort I put in I will be able to place bets and not have to worry about it, I was just looking for a hint because I've been trying everything that has come to mind and the only one method that works, only works for the one game and throws the others completely out of wack

[Pokerjoe]

Actually, eliminating biases is a great way to improve a model. There may be times when a bias is okay (such as if your model is biased toward centrist results when facing outliers in NFL totals), but generally they're a sign that you're off target, and your algorithm's error is only being revealed when sufficiently magnified by outlier challenges.

Boy that's a bad sentence. I mean, if a league's average total is, say, 200, and your model is off by a 1/2 point at totals around that number, you won't see it. But by the time you get to totals of 220 or 180, that error might be magnified enough that you do see it, and it won't be neutral and thus invisible, instead it'll be consistent, large and thus obvious, as in your case with high NBA totals. So, it's a good thing. Not as good as not having the bias, but better than having a bias you haven't detected.

[the shadow]

Totals are derived from comparing each teams avg. score per game to the league avg.
In your case, GS avg. is 208 per game(8 over the league avg of 200)
SA avg. is 205 per game(5 over league avg.)
Thus 8+5=13
Take league avg(200+13=213) .This is close to posted total of 214.5
You can do this with all games and find that this method will be within 2 or 3 points of the posted total.
This is the method the books use to create totals, then adjust slightly for injuries,schedule etc.

[the shadow]

It's funny, the lal-hou game should of had a total of 219 using above method. It closed at 205. REASON= lal went under 7 games in a row and hou went under 4 in a row. The books adjusted accordingly. We still have to handicap!