Alfa1234, I think you and I will have to disagree that anything under 10,000 wagers is not a valid sample. Since only 15-33% of games are selected as wagers that means you would have to analyze 30,000 to over 60,000 games to be valid per your standard. First, some edges don't last that long. Second, that would be far too many seasons and things do change over time. I hope you don't wait for 10,000 wagers for validation before risking serious money.
I have used the very concept that I am describing in this thread successfully for over 40 years. While I do not have 10,000 NBA wagers I do have thousands of NBA wagers (with a success rate well above 3 std dev) and across all my wagering over the years way over 10,000 wagers using the very same concept profitably in various other sports. I wagered very part-time until 2011 when I retired (10 years early due in part to gambling winnings) and have been wagering almost daily since I retired. I realize what I wrote may generate some questions but this is all I feel comfortable writing about this at this time.
As to your theory, I have not beat the closing line enough of of the time or by enough margin to generate profits by that method and I keep meticulous records. Which is also the reason for my confidence in my results. My profits come from the fact that the sub-set of games that I wager on win at a rate significantly higher than 50/50 and that is what produces the profits.
As a side note Kelly is a wonderful thing. I read a lot naysayers on these and other boards that try to discredit Kelly but frankly it is mathematical genius. First, to use Kelly you must have an edge and unfortunately that means that most people cannot use Kelly. It also seems to me that a lot of the naysayers haven't ever used Kelly and/or don't understand its application. If you have an edge that can be quantified correctly (a lot can be written about this alone but I don't have the time right now) I suggest that you seriously give Kelly a chance. My recommendation would be to use half-Kelly for added safety and less wild swings of bankroll. Going back to validation, for me it really does't matter, I have already replaced my entire bankroll with profits this NBA season so even if I lose my entire bankroll at this point, I will not have lost a nickel. I do this year after year. I debated whether I should write the following but here goes: My half-Kelly wagers are as follows:
# of wagers to make
1 9% of bankroll
2 8.2% of bankroll on each wager or 16.4% of bankroll at risk
3 7.5% of bankroll on each wager or 22.5% of bankroll at risk
4 6.8% of bankroll on each wager or 27.2% of bankroll at risk
5 6.2% of bankroll on each wager or 31.0% of bankroll at risk
6 5.6% of bankroll on each wager or 33.6% of bankroll at risk (The most wagers I have had in one day)
I'll stop here as I am sure that many of you will find this either unbelievable or too risky or?
Joe.