#1
|
|||
|
|||
Anyone here knowledgable about PokerStove
I've been running some simulations. My current interest lies in what cards have pot equity vs. multiple limpers or cold-calling after 2-3 players entered the pot. Unfortunately, there are just too many permutations to examine all the opponent combo's with 3 random opponent hands.
The question is... what is the general algorith for exploring opponent hands? For example, i might pick 1 player with JcTd and wish to run 3 opponents with random hands. After about an hour, i've reached 1% of all possible hands, and the pot equity begins to converge. However, this data is obviously meaningless if the first cards the program runs through assumes the first card is clubs for each opponent. I know if i wrote the code, i'd make the probability of the any suited card in each opponent's hand random conditional on the fact that the non-random hand is already holding a club and a diamond ( ie: probability random hand holds a club = 12/50 + 12/49). as well as distribution given the card value. Fair assumption? Also, do you think the 1% mark after 1 hour of analysis is enough time to get a reasonable 'rough idea' of how a hand performs vs. 3 random opponents. ie: can i call 2 limpers from LP when i think either the small blind will raise and the BB folds or the SB folds, giving me 3 opponents. |
#2
|
|||
|
|||
Re: Anyone here knowledgable about PokerStove
[ QUOTE ]
I've been running some simulations. My current interest lies in what cards have pot equity vs. multiple limpers or cold-calling after 2-3 players entered the pot. Unfortunately, there are just too many permutations to examine all the opponent combo's with 3 random opponent hands. The question is... what is the general algorith for exploring opponent hands? For example, i might pick 1 player with JcTd and wish to run 3 opponents with random hands. After about an hour, i've reached 1% of all possible hands, and the pot equity begins to converge. However, this data is obviously meaningless if the first cards the program runs through assumes the first card is clubs for each opponent. I know if i wrote the code, i'd make the probability of the any suited card in each opponent's hand random conditional on the fact that the non-random hand is already holding a club and a diamond ( ie: probability random hand holds a club = 12/50 + 12/49). as well as distribution given the card value. Fair assumption? Also, do you think the 1% mark after 1 hour of analysis is enough time to get a reasonable 'rough idea' of how a hand performs vs. 3 random opponents. ie: can i call 2 limpers from LP when i think either the small blind will raise and the BB folds or the SB folds, giving me 3 opponents. [/ QUOTE ] A lot of the goons in the 1TT forum use Poker Stove. I might give a try over there. |
#3
|
|||
|
|||
Re: Anyone here knowledgable about PokerStove
If the problem is too complex, just use Monte Carlo approximation. It converges fast enough, doesn't it?
-pix |
#4
|
|||
|
|||
Re: Anyone here knowledgable about PokerStove
try a pm to andrew prock
cheers |
|
|