[roasthawg]

Originally Posted by Justin7

To approximate for games near 50%:

Take square root of sample size, and divide by 2. For 100, this would be 5. Every 5 above 50% would be on deviation.

If your system went 60-40, that would be 2 standard deviations, or about 98% confidence.

On another note, if you're "data fishing", you should expect to find one system like this for every 50 you check, even if there is no logic to it. The more crap you search, the more crap you find.

In addition to having good statistical results, you need to have a logical basis why this system is better than flipping coins. And if your idea is simple, it's a fluke (other people would already know it, and the market would correct for it).

so if you're system went 60-40 are you saying you would use 55% for your expected win%? thanks for your help...