Is there a formula to figure out the confidence intervals in DONs like there is in cash games. For example If you play 40,000 hands of limit holdem with a standard deviation of 20, your winrate would have a 95% chance that your true winrate is within +/- 2 BB of your current average. (We'll just assume game conditions, your play, etc. always stay the same) So in double or nothings, say you win 55% of your tourneys over a sample, could be 100 games, 1,000 games, 10,000 games, etc. How would you know how accurate that percentage is, i.e. you have a 95% chance of being within 3% of 55% (range 52%-58%) or 5% or 10% or whatever.

Bonus question: Does anyone know the standard deviation in DONs? I would figure that due to the nature of them, it's going to be pretty much the same for everyone. Thanks.

http://en.wikipedia.org/wiki/Binomial_distribution

If the probability you win a DON is p, then the SD of 1 DON is sqrt(p*(1-p)). The SD of n DONs is sqrt(n*p*(1-p)). If n*p > 10 and n*(1-p) > 10, you can use a normal approximation to determine a confidence interval.

To add: The SD won't be the same for everyone. It's dependent on the probability of winning. For example, a player who wins every time will have a SD of 0; the same goes for a player who loses every time. A mediocre player who wins half the time will have the highest SD (the SD of a binomial distribution is maximized when p = 0.5).