Confidence - Win Rate

[CMangano]

I am wondering what a good confidence level for your win rate is. I have Stat King and "the Professor" seems to think that 67% is the level you want to be at. Whenever I ask if my win rate is accurate, he says "...at 67% confidence..." I am at 66.6% confidence right now. What confidence level do most of you want to achieve before you feel that your win rate is accurate?

[Copernicus]

Do you separate your win rate by stakes? Do you always play against the same number of opponents? Do those opponents have identical standards for their play?

If you answered no to any of the above and try to maintain a single average win rate, the answer is never.

Do you enjoy playing? Do you play within your means? Are you a non-professional player? If the answer to all of these is yes, who cares whether you know your win rate(s) to a 1 2 or 3 standard deviation confidence? Its mental masturbation anyway.

Are you a professional player? Does your sustaining your career depend on not playing at too high a limit for your bankroll? Do you maintain detailed records of your wins/losses under different playing circumstances? Do you maintain a "book" on players you face regularly? If the answer to all of the above is yes, you don't NEED to know how accurate your win rate is, you know without the math whether you are playing at the right level or not.

I would be shocked to hear that any significant number of the biggest names in the game (other than Mason or David) have the vaguest idea what their "win rate" is.

[BruceZ]

See my response to your post in the standard deviation thread on the probability forum. A player can easily overestimate his abilities, and many "top players" don't play nearly as well as they think they do. Much money is made off of those who think they play well. Many pros have gone broke due to lack of understanding of swings and bankroll requirements. If being a professional meant that you know you are good enough to sustain your bankroll at a certain level, then this wouldn't happen nearly as often as it does. A good player who is likely to go broke may not be any better off than the guy shooting craps. Conversely, it is also possible to think you are a bad player when your results can be explained in terms of expected swings. Intuition about results can be very misleading. Those who don't keep stats don't do so out of ignorance, laziness, or fear about what the results may say, but never as a result of being a good player.

On the other hand, I support separating your data into different classes depending on known variations caused by different limits, players, aggressive vs. loose games, etc. Then do statistics on each set of data.

In answer to the original question, I don't know what StatKing says, but 68% confidence is arbitrary. It just corresponds to 1 standard deviation. It means that your results would lie within this range 68% of the time with the given assumptions about your win rate and standard deviation. You can make confidence intervals for any confidence level. The 95% confidence interval is twice as big as the 68% confidence interval.

[SittingBull]

division(game , structure, Etc. will average a win-rate session of about 65% with a confidence level of at least 90%.
HappyPokering, /images/graemlins/smile.gif
SittingBull