would anyone test drive this, or would I be wasting my time?

[eastbay]

With all the "all-in or fold" analysis that has been done, I thought it would be interesting to field test some of the simulation generated "optimal" strategies to see how they fare against real players rather than simplistic counter strategies based on fixed hand rank lists or similar.

So I was thinking about putting a CGI program up on a web page where you (anyone who wanted to try it) could duke it out with a computer program heads-up, seeing if they can beat the "optimal" strategies generated by the preceeding simulations. By starting the chips at 5000 apiece and blinds of 250/500 (which isn't an at all unrealistic PP scenario), the fact that you can only all-in or fold shouldn't be considered absurdly artificial.

Would anyone have the patience to run enough of these out to get statistically significant resuls, or would that be too boring with no direct $$$ incentive? Since the computer would act instantly and I wouldn't bother with fancy animated graphics, the games would move as fast as you wanted them to, so you could conceivably crank out tournaments pretty quickly.

If a few of you thought you'd be willing to try it, it might be worth writing the thing, which would probably take a significant time investment (a few tens of hours). What we would get out of it is a sense of the applicability of these all-in or fold simulation results against real players. There might be significant endgame equity in here. Who knows.

Let me know. Interested?

In case you're wondering, I do try this stuff out "live" but "live" is always less than ideal. First off you don't get it heads-up at the end every game, and when you do, the stacks are often so skewed that clearly the right strategy is altered by that fact. Of course, eventually we'd want to work the short-stack and big-stack problems as well. But you have to start somewhere.

Working the even-stack problem systematically is probably a good place to start.

eastbay

[LetsRock]

If I was building something like this, I'd set the program up to run the simulation thousands of times each way. The whole point of computer simulation is to eliminate the tediousness of manually doing the same thing over and over.

Set it up so you can input the parameters (your hand, stack sizes, antes, opponent's hand if you like) and let the computer go to town. Have it reshuffle the unknown cards each time so you get a random result.

[eastbay]

[ QUOTE ]
If I was building something like this, I'd set the program up to run the simulation thousands of times each way. The whole point of computer simulation is to eliminate the tediousness of manually doing the same thing over and over.


[/ QUOTE ]

That's already been done. Now the question is: how good are the resulting optimal strategies against real players?

eastbay

[ctv1116]

I'd certainly be willing to test it out. Certainly something to do in the early stages of Party SnGs.