What is wrong with Card Player's Odds Calculator?

[Cohiba Al]

I've noticed that entering in two player's starting hands and calculating the preflop odds will often give slightly different answers each time you calculate them (the same two hands). Often the difference is off by .4% which might not seem like much. I was wondering why the calculator would even vary at all? How does a binary computer making a mathematical computation come up with different answers each time? Is their free calculator worth using?

[gaming_mouse]

I'd guess that it is using a Monte Carlo simulation.

Not sure, cause I user pokerstove instead, which also has the option of a complete enumeration, and with this method the results never vary.

gm

[fnord_too]

They do run a monte carlo, which means that the answer is only a resonable approximation, but not the absolute answer. You can get an exact answer (I have some code that will calculate it) but the estimate is good enough for most purposes.

[gaming_mouse]

[ QUOTE ]
(I have some code that will calculate it)

[/ QUOTE ]

No need for custom code in this case. Just download pokerstove (free):

www.pokerstove.com

[Gamer]

What exactly is a mote carlo simulation? Do they just play lots and lots of hands and registering how many time you would win, or is it somewhat more complicated?

[gaming_mouse]

[ QUOTE ]
What exactly is a mote carlo simulation? Do they just play lots and lots of hands and registering how many time you would win,

[/ QUOTE ]

yes

[MortalWombatDotCom]

[ QUOTE ]
They do run a monte carlo, which means that the answer is only a resonable approximation, but not the absolute answer. You can get an exact answer (I have some code that will calculate it) but the estimate is good enough for most purposes.

[/ QUOTE ]

i am shocked and appalled. before i knew there was free software out there to do this, i wrote my own. i planned on writing it the straightforward way (enumerating all the hands with no optimization of any kind), then seeing how fast it ran and trying to optimize it in some terribly clever way. well, i wrote it and it took about 3 seconds on my laptop to compute the answer (slightly more for each additional hand contesting the pot). so i didn't bother to make it faster.

anyway, my point is, why on earth would somebody use monte carlo simulation when the exact solution is so cheap to calculate?

BTW, i have never used any of this free software, so i don't know exactly what it lets you do... is any of it open source?

[droidboy]

[ QUOTE ]

BTW, i have never used any of this free software, so i don't know exactly what it lets you do... is any of it open source?

[/ QUOTE ]

Check out:

http://pokersource.sourceforge.net/

- Andrew

[BlueBear]

[ QUOTE ]
anyway, my point is, why on earth would somebody use monte carlo simulation when the exact solution is so cheap to calculate?

[/ QUOTE ]

Very simple, if you need to expensive the performance of your hand against a few of your opponents unspecified random hand (say AA against 5 random hands without a flop specified), the gain in computational time is huge when compared to a complete enumeration.

[MortalWombatDotCom]

[ QUOTE ]
Very simple, if you need to expensive the performance of your hand against a few of your opponents unspecified random hand (say AA against 5 random hands without a flop specified), the gain in computational time is huge when compared to a complete enumeration.

[/ QUOTE ]

i didn't mean to suggest that there were no jobs for which monte carlo simulation is the appropriate tool. i did mean to suggest that the stated problem (matching a small number of specified hands against each other) is not one of those jobs. *shrug*