stats help

[Guest muppet]

what is the best way to tell if a system is statistically significant - apart from looking at financial gain and odds? eg over / under total goals in a game? assume prob of "under" = 0.5 assume prob of "over" = 0.5 out of say, 200 games or trials, how significant is eg 150 successes or 150 correct predictions? i know the mean is 200x0.5=100, but how significant is 150? do you use the binomial distribution? is there a critical value? i've seen it on excel functions but unsure about CRITBINOM. can anyone explain the best way to see how many i need to be winning, assuming odds of 0.5? hope this is clear. any bright ideas? will be obvious to someone. cheers in advance. :\

[Guest OddsAgainst]

Re: stats help There are two methods commonly used, the one-tailed t-test and the chi-test. Let's look at how to calculate the t statistic. Assuming you have the bets in question in a spreadsheet and a column for the profit / loss for each bet. (note: profit, not return) First calculate the average win. (Obviously if the bets are showing a loss at then this calculation is pointless). Then calculate the standard deviation of the average win. Now calculate the standard error which is the standard deviation divided by the square root of the number of bets. Finally, divide the average win by the standard error, the result is the t-statistic. The bigger the number the more confident you can be that your results are down to skill rather than luck. With a t value of 3 you can be 99% sure that your success is down to skill, a value of 2 95% sure and at 1 80%. To get the intermediate values you need to have the probabililty distribution which I don't have. Anyway, that is the theory if you are still awake. I will refrain from giving an opinion on the validity of the method as I know that others will come along soon with some erudite opinion. ;)