View Full Version : Statistics Project

11-01-2004, 02:18 AM
It's that time of year in statistics class. Time for a project to show that you understand all the material so far. I want to do something poker-related, but I don't want it to be a free lesson, even to a class of 60 some-odd people. That being said, I think a two-proportion test of flops seen and money won would be out of the question.
Obviously there are millions of hand histories between the people on this site, so my sample size is huge. Does anyone have any ideas of a poker-related statistics problem that wouldn't involve me teaching the class how to play better? It doesn't have to be serious either. One person in a previous class compared how alcohol choice at Duke depended on gender. Turns out being a woman did make the person more likely to drink liquor rather than beer, and that in either case, Duke still sucks.


11-01-2004, 12:38 PM
I have a fascinating idea that I hope that you will consider. Leave pokertracker on, observe 20 3/6 games at one time, collect all the data of the players and hands.

Then looking at the summary, do a statistical study (using correlation methods & hypothesis testing) to these test these hypothesis. Some simpler ideas may include:

1) is there a evidence of a relationship between the preflop tightness and the win rate of a player?
2) evidence of a relationship the preflop raise (aggressiveness) and the win rate?

These may involve more work. You could also see if

1) players are more aggressive in the night time compared to day time?
2) are players more loose at the night?
3) what about during the weekend?

Finally, If you into some simple programming, you could examine what the effects of maniacs (PFR>50%) are on the rest of the table. Do players typically become more tight? or do they become more maniacal? It would be a very interesting study.

These are just some ideas to get you started.