Attention Programmers:

[Jman28]

I thought of what seems to be not too difficult of a program to make that would help my SnG play a good deal... Especially with multi-tabling.

Could you make a program that reads my preflop cards and displays their odds against probable hands, and perhaps in a color code.

For instance, the following table shows the odds I'm talking about. I'd say for most SnG play, toward the end you're looking at the 25% or 40% columns for getting called on a push.

http://home.earthlink.net/~craighowald/data/matchup2.html

I've been told this table is both copyrighted and inaccurate, but it's just the example here.

Is this difficult? Was I clear?

-Jman28

[ChrisV]

If by "reads your preflop cards" you mean scans them off Party Poker, that's pretty difficult.

I've been thinking for a while about writing a program to run simulations for heads-up at the end of a SNG and discover the optimal game-theory solutions for both players at various stack sizes.... I'm not sure I'd actually release the results though :P

[CT11]

[ QUOTE ]
If by "reads your preflop cards" you mean scans them off Party Poker, that's pretty difficult.

I've been thinking for a while about writing a program to run simulations for heads-up at the end of a SNG and discover the optimal game-theory solutions for both players at various stack sizes.... I'm not sure I'd actually release the results though :P

[/ QUOTE ]

It's not that bad really. It is how ever against party TOS which is why I personally _wont_ do this.

There have been many warnings about using computer simulations to come up with "optimal" play. Remember poker cannot be "solved" like chess, so there is no optimal "game-theory" _solution_. Also assuming your opponent has random cards and taking every possible combination of actions and seeing which is best is also wrong.

Basically be careful with what information you take from a simulation.

~CT

[Chief911]

Why not learn how to play poker instead? Novel.

[ilya]

Why in the world would someone do this for you? Don't you think you are being fantastically presumptous?

[eastbay]

[ QUOTE ]

There have been many warnings about using computer simulations to come up with "optimal" play. Remember poker cannot be "solved" like chess, so there is no optimal "game-theory" _solution_.

[/ QUOTE ]

Wrong.

eastbay

[eastbay]

Not that difficult. Maybe 20 hours of work. If it weren't against party TOS, I might do it for $75/hr.

eastbay

[Paragon]

You can actually write a pretty simple program to read your cards from the hand history file on your hard drive. I'm not sure how that would be against the TOS as I haven't looked through that. As a comp sci/math major I've entertained the idea of making my own pokertracker-esque thing that simply scans through these files to get some stats for my own use. It'd be a little different to do something in real time, but it would definitely work. In any case, I'm not willing to do it myself /images/graemlins/smile.gif

[Jman28]

[ QUOTE ]
If it weren't against party TOS

[/ QUOTE ]
Yeah. Good point. I didn't think of that.

[Jman28]

[ QUOTE ]
Why in the world would someone do this for you? Don't you think you are being fantastically presumptous?

[/ QUOTE ]
I was thinking that it would be a program that a lot of others could use as well.

[ChrisV]

"There have been many warnings about using computer simulations to come up with "optimal" play. Remember poker cannot be "solved" like chess, so there is no optimal "game-theory" _solution_. Also assuming your opponent has random cards and taking every possible combination of actions and seeing which is best is also wrong."

Actually this is incorrect on both counts. Heads-up holdem is a zero sum game, so there exists a (very complicated, incredibly difficult to derive) optimal game theory solution. A player playing this solution cannot be defeated - the best you can do is break even by playing the same strategy. However this strategy will be non-optimal vs a lot of opponents (it will win, but not as much as could be won).

I'm talking only about heads-up at the end of an SNG, with blinds of 200/400 and up. This allows me to vastly simplify the problem by assuming that the only two possible actions are allin and fold. I also get to assume the opponent's hand is random when hes in the BB. Basically I'll plot every possible strategy for one player against every possible strategy for the other player and find the point at which both players are maximising their potential. I'll then look at how this optimal strategy does against every other possible strategy from that player, to make sure it isn't sacrificing stacks of earn against some guys.

[eastbay]

[ QUOTE ]
"There have been many warnings about using computer simulations to come up with "optimal" play. Remember poker cannot be "solved" like chess, so there is no optimal "game-theory" _solution_. Also assuming your opponent has random cards and taking every possible combination of actions and seeing which is best is also wrong."

Actually this is incorrect on both counts. Heads-up holdem is a zero sum game, so there exists a (very complicated, incredibly difficult to derive) optimal game theory solution. A player playing this solution cannot be defeated - the best you can do is break even by playing the same strategy. However this strategy will be non-optimal vs a lot of opponents (it will win, but not as much as could be won).

I'm talking only about heads-up at the end of an SNG, with blinds of 200/400 and up. This allows me to vastly simplify the problem by assuming that the only two possible actions are allin and fold. I also get to assume the opponent's hand is random when hes in the BB. Basically I'll plot every possible strategy for one player against every possible strategy for the other player and find the point at which both players are maximising their potential. I'll then look at how this optimal strategy does against every other possible strategy from that player, to make sure it isn't sacrificing stacks of earn against some guys.

[/ QUOTE ]

I did this a long time ago. The posts may still be in the archives.

You'll have to limit your solution space quite dramatically, but it hardly matters for the answers.

Be prepared to be disappointed by the answers. Basically, unless one of you folds too much, it's almost entirely a crapshoot.

eastbay

[ChrisV]

I'm well aware of that.

I just looked up what the 70/30 rule you postulated means in practical terms and it turns out to be more or less exactly my usual strategy /images/graemlins/smile.gif

[eastbay]

[ QUOTE ]
I'm well aware of that.

I just looked up what the 70/30 rule you postulated means in practical terms and it turns out to be more or less exactly my usual strategy /images/graemlins/smile.gif

[/ QUOTE ]

However, if you're a programmer who is interested in such things: wanna help me work the 3-way problem?

eastbay

[ChrisV]

Not really /images/graemlins/tongue.gif

It's considerably more complicated and there may not be an optimal solution in any case. I program for a living so I'm not that keen to do it in my spare time as well.

[eastbay]

[ QUOTE ]
Not really /images/graemlins/tongue.gif

It's considerably more complicated and there may not be an optimal solution in any case. I program for a living so I'm not that keen to do it in my spare time as well.

[/ QUOTE ]

Why do you think I'm asking for help...

However, there is at least one equilibrium triple. I believe that's guaranteed by Nash's Theorem.

eastbay