Re: Hold\'em simulator (some competition for PokerStove)
 

[ QUOTE ]
I ran a simulation with AA-AT,KK-22,A*s vs 4 random hands.
I then rearranged it as AK-AT,AA-22,A*s vs 4 random hands.

These ranges are equivalent but the results were minutely different. It's probably that they didn't converge at 600,000 hands but you may want to look at it.

[/ QUOTE ]


Thanks for the feedback. Yes, as you supposed, what is happening is that a new monte carlo simulation is being run, which (as you would expect) gives slightly different results.

I haven't yet decided 100% what I want to do, but I may do any or all of the following in the future:

1) Allow for "re-run" to increase sampling
2) Make the monte-carlo stuff faster
3) Convert all ranges to equivalence classes (so that your two examples would resolve to the same database entry, and give the same result).

I'll see what demand there is. My feeling is that in practice, 600,000 hands is plenty and practically guarantees results within half a percentage point. The pure geek-value of convergence has its attractions, but on the other hand, there is value in having the simulation result links be idempotent, as I will probably be using the simulator in other parts of my website and it probably looks weird to have values change on you.

Cheers,
bachfan