I wondering how you calculate the risk of just being lucky. I guess this http://www.fourmilab.ch/rpkp/experim...sis/zCalc.html formula works for if you played to the odds of 2. But say you played with ML odds to 3 and hit 51-49 after 100 bets. How do you calculate the risk of just being lucky to be on the positive side?
z-score above 2 is OK, above 3 is good, above 4 is excellent, but none are any absolute guarantee you will win going forward. BTW, calcing a z-score with the info you provided is impossible. z-score calcs against point lines is very straightforward, MLs much more difficult.
Bottom line, with the info you have provided, it's luck: so bet absolutely minimal bets until your running z-score (moving average of last 100 games or so) at least gets above 1.5 or so.
Z-score gives a measure of consistency.
Action points give a measure of strength of picks.
A combination of both may be best.
For a short term record, action points will paint a pretty reliable picture of luck/lack of luck. A capper can be 42-33 (56%) with 285 action points, and another 47-34 (58%) with 30 action points. As long as the record is for the same league, it should easily point out the luckier player. The capper up 13 units has only 1.4 action points for each of those excess wins. The other player, up 9 units, has 31.7 action points per excess win.
Why can't I remember names of books? I know it dealt with randomness. An example in the book was about running the stats from the entire history of MLB and simulating 100+ years of ball games. Based purely on stats and randomness, it was calculated that there should have been 11 hit streaks as long as Dimaggio's. Not to slight Joey D, but someone had to do it.
Like Maris hitting 60 homeruns. Someone had to do it.
Like Hollywood movie producers. How do you know which ones are good, and which ones are getting lucky, or the good side of the random draw?
And Fund managers. Who knows the market and is making money, and who is lucky?
And following amateur cappers on forums. If they start bad, they will probably stop posting. If they start good, they will continue posting. But how do you know if they are good, or just purely distribution in action?
I would like to see a scientific bet sizing chart, per league, for sides and totals, for 50-99 bets, 100-249 bets, 250-499 bets, and 500+ bets, where the suggested bet size is based on 'action points per excess win'.
Example:
NBA sides, 50-99 bets, action points per excess win: 15-16 pts => bet size
NFL totals, 100-249 bets, action points per excess win: 3-4 pts => bet size
To most players, who can't accurately determine their edge, this could be a lot more helpful than Kelly.
Sample size is the most significant variable in any such calculation. If we are talking about 10-20 wagers, any type of calculation is pretty much useless.
If you take all the touts in the world, of course some of them will get lucky and do well. You can run stats on their plays that will tell you it's highly unlikely their streak is just 'luck', but out of all the touts out there......you see what I mean anyway.