Quote:
Originally Posted by EazyDoesIt
What exactly do you mean by: "Builds a model of your typical opponent"
For any given statistic, S, you could plot a graph showing what fraction of players you've observed have an observed value of S less than x. Like the red line in the graph for WTSD below.
Unfortunately, for many players we only have a few observations to work with, so the observed distribution has much higher variance than the true distribution. You can see, for example, the abrupt vertical spike at x=50%, where many players were observed on two flops and went to the showdown once. There are other noticeable spikes at x = 0%, 33%, 66%, and 100%.
Via the application of calculus, Poker Sleuth is able to remove the extra variance created by the inaccurate observations to come up with a model for the true distribution of WTSD among players, shown as the green line in the graph below.
Poker Sleuth uses the model as the Bayesian prior distribution when computing credible intervals. A bit more is described
here.
Quote:
Originally Posted by EazyDoesIt
One very important question: Have you contacted the poker sites to make sure all the futures in your program are legal at there site, so your customers don't get in trouble?
Not yet, but I have read over their policies carefully.
Quote:
Originally Posted by EazyDoesIt
This looks promising, especially the interval for statistics. I do hope you can make the performance efficient.
Performance will definitely not be a problem.
Quote:
Originally Posted by EazyDoesIt
Even though I don't play Omaha, I would recommend supporting it. Because that would give you a big edge on the competition and could help a lot putting your product out there. Right now Omaha players don't have any good options.
Good observation. Thanks!