Bet against a model with highly negative hit percentage?

**FourLengthsClear** · 12-17-11, 07:44 PM

Your model is badly flawed if you are having that sort of differential. You should never be seeing an 8 point differential.

**Lockitup1x** · 12-17-11, 09:59 PM

No.

**roasthawg** · 12-18-11, 01:11 AM

In all likelihood it's a bad model... don't bet it either way.

**InTheRed** · 12-18-11, 01:26 AM

I disagree with everyone here.

Assuming you have extensive backtesting and you have over 1000 system plays, if it is consistently hitting at those ratios, i would bet the big ones and fade the small ones.

In my O/u NBA system, I am accounting for many different variables in pt differentials and scenarios. If they are hitting, they're hitting.

**u21c3f6** · 12-18-11, 09:03 AM

Originally posted by strixee

I have back-tested my model and found when my line differs by 8+ points from Pinnacle line, I'd be hitting at least 55%.
What's more interesting however, is that if I'm within 4 points from Pinny, I'd be hitting under 45%. Is it wise not only bet when I have an edge, but also fade my model when my "edge" is clearly negative?

Absolutely, assuming it is not data-mined.

Start small to confirm your results and increase (not more than Kelly) as your results prove profitable.

Joe.

**Wrecktangle** · 12-18-11, 10:13 AM

I agree with the model faders; not the same as fading a human who will change criteria to try to win.

But you have more work to do, you need to figure out what is the root cause and get it positive. Running "Black Boxes" will crash your BR eventually. You must understand what factors are causing the problem to better understand the stats behind the sport.

**strixee** · 12-18-11, 04:16 PM

But you have more work to do, you need to figure out what is the root cause and get it positive. Running "Black Boxes" will crash your BR eventually. You must understand what factors are causing the problem to better understand the stats behind the sport.

Good answer, I'll be thinking more about the model.