Is normalizing alone adequete for small markets?

**Justin7** · 03-05-09, 03:35 PM

Originally posted by Wayfarer

For small markets where the lines should be the softest would simply normalzing a number of important offensive and defensive statistics be enough for identifying edges?

It seems to be working for me so far but because of my small sample size I am not sure whether it is just randomness currently working in my favour or an actual edge.

It really depends on how efficient the market is. You can beat a small market just with basic stats and normalization, but the more work you do, the better your results will be.

I would start with a stat based approach, learn a bit about the sport, and try to isolate two things: instances of luck that skew statistics, and special situations where pats results are less predictive. If you can do all this, you can destroy smaller sports.

**Arilou** · 03-06-09, 01:43 PM

There are a number of steps you have to go through in order to properly go from statistics to odds. Normalizing the odds is a great first step, and controlling for major changes or non-predictive events is a great second step. The key is translating those fixed stats back into odds of winning, or a total or point spread, and that can be hard especially if you're not an expert in the sport.

However the approach is excellent, as you are attacking where the lines are soft. If your model is solid, even if it would be no better than random in the major sports, you should still win easily.