[JOHNPRUSSELL]

Justin: in order to test a probability theory, roughly how much of a poulation size (N) do you think you need in order to get accurate results? I think you once said around a 1,000, is this correct?

[Justin7]

That's a complex question. What are you testing, and what is your error tolerance?

[JOHNPRUSSELL]

Well, its obvious that i dont know much about statistics. I took one course in college and cannot remember much. Therefore as embarassing as it is, i dont even know how to answer your question. But I am going to tell you what I have done, currently doing, and hopefully you can give me a little expertise as to how to do it better, or at least let me know if i am on a collision course for failure. So far I havent bet any basketball, so i havent lost anything. I picked Auburn to win the BCS at 15/1 about the middle of October so I have a decent bank roll. I am trying to play a little smarter. My past plays have been oriented around: a)catching the "steam" early and getting it down on slow movers. b)correlated pars (getting tougher to get down), c)playing dogs late d) playing 3's on pro football.

Here is what I have and here is what I am trying to do. First, I have on EXCEL every ncaab game in rotation order since the first of the year(1/01/2011). I have the open/close line/total, half time score,line/total, final score. this amounts to about 2500 games.

The first thing i did was to try to analyze front end favorites of -10 or more that were losing at the half, and then play them to cover the 2nd half. Out of 2500 games, about 300 had favorites of -10 or more losing at the half. After charting 300 games, it turned up that 154 front end favorites covered the 2nd half and 146 dogs covered the 2nd half. Obviously you see this was a waste.

Now, I am looking at playing dog/under pars where the favorite is -10 or more. Of my 2500 games, about 400 fit the "test" group. This seems to be working pretty well out of 400 games, 128 fell dog/under 5 had a push with a winner, and 267 were losers.

Thus, in order to tell if my dog/under theory is playble, how many games would i need to chart in order to feel reliable enough to play at a profit? Do you know of a better way to test a hypothesis? Any advice you give will be greatly appreciated.