I have to do a statistics project for my math class, and I chose as a subject Poker. I have a lot of interest in the game and see a lot of potential for a project since there is so much math involved. Now my problem is to come up with a question which I can answer using a simulation or by playing many, many hands. The only thing that I’ve come up with so far is whether it is better, when given a hand like Jd, Jh, Kd, 7s, 3c to throw out the king, 7and 4 or to throw out only the 7 and 3. This is problematic, however, because how can you give a value to having two Jacks and two Kings to having three Jacks?
Answer 1:
That’s not a good candidate for simulation. That question can easily be addressed by direct computation. I think the issue of giving value to two pair or three jacks in draw poker was addressed in an article in Operations Research. It didn’t use much advanced math either. You can find the citation by going to the IFORMS website and doing a bibliography search on keyword poker. That article did not use simulation. You might think about 7-card stud for questions that get too cumbersome to address directly and are better suited to simulation.
Answer 2:
You could try a problem like if I am sitting in an O/8 game with an A-5 and a 2-3-8 flop, what is the likelihood that another player has seen the flop with an A-4 or an A-5. The probability varies with the number of players calling pre-flop. This would vary with how loose the game was, starting hand selection, etc. The Turbo O/8 program would be useful to address this. The answer would be useful to me the next time that I am seated in a game with the A-5 and a 2-3-x(x=6+) flop. A different type of problem would be to simulate the probabilistic distribution of outcomes in a session (or week, month, or year) for a poor player, average or good making some assumptions about typical pot size, contribution, and probability of winning hands. I’d like a curve that shows how likely is it that I’d lose 8 sessions out of 10 if I average winning 2BBs per hour. I want to be able to see when I should begin to disbelieve that I really average 2BBs.
Answer 3:
Since this evidently is a statistics project, here is my suggestion: calculate the probability of various hands being dealt in 5 cards (as they are in 5-card stud). I suggest 5-card stud because the calculations will be the easiest. Also, you can check your numbers (remember to show your work, now) for the probability of hands against any number of poker sources. Then simulate a buttload of 5-card hands. You can compute standard deviations, variances, and other interesting things for various numbers of
hands, and get some real-life, hands-on experience in how different theory and reality can be. How likely is it that the hands will reflect probability calculations after 100 hands are dealt? After 1,000? (If you have a computer, after 10,000 or 100,000?) You can begin to see why even good poker players can lose a lot of money, especially in the short run. Notice how, for large numbers of hands, the results tend to “converge” around the probable outcomes. This is a demonstration of Large Numbers.