[noober]

I was able to find a 10% EV strategy but the test was done on a sample of 260. I know that's not very big and since I am not very good at statistics can someone tell me if i should trust this strategy. Thank you.

[TomG]

size doesn't matter

[ljump12]

Size absolutely matters, and 260 games is *probably* not enough to be really confident. First you are probably "data snooping" to find that trend. What you really need to do is split you backtesting into a training and test set, but for that you need a decent amount of data. I'd need to know more about how you backtested the trend, what sport, what odds your betting at etc. to tell if you can really trust it.

Knowing nothing else, I'd say you should be very skeptical that it will perform +10% in the future.

[noober]

I am betting on tennis. The data set is 8000 matches. Out of those matches, the strategy that I tested is applicable to only 264. Out of these 264 matches it won 136 times. The odds are 2.15 so I concluded that this strategy has EV of about 10%.

Thank you for the help.

Thanks a lot for the Intro to Research Thread as well.

[Wrecktangle]

The deal with sample sizes in sports is, frequently the sport changes (rules, coaching trends, etc) before you get enough data to really statistically prove you "have enough data." The NFL is my best example of this. Next, the typical test used to show you have enough data is typically based on Gaussian distributions, which few sports distributions are.

Not knowing much about what exactly you are doing, I'd say a sample size of 260 is enough, but don't be surprised when your method "goes south" on you so be defensive in your betting.