Poker Probe: Crude Analysis

[ninthyaga]

Using Poker Probe for some crude analysis. Found that AA will win a little over 50% of the time against 4 other random holdings (given nobody folds, everyone shows down). Assuming no folding - let's say everyone is all-in for the same amount - this number would still be off from real life because the four hands that would stay in with you would be most likely better than average. So the Poker Probe is actually overestimating the value of the main hand. How many more random hands do you think I have to add to compensate for this inaccuracy?

[Phat Mack]

I've thought about the same thing and have decided that it depends on the table you wish to simulate. If you want to simulate AA against four opponents at a table where they play the top 50% of their hands, you could try running against 8 opponents. Four of the eight opponents would have hands in the top 50%. The problem with this approach, however, is that the four with the bottom 50% stay to the end -- and can suck out. That'll skew AA's EV downward.

There's a freeware program that allows you to put opponents on ranges of hands. It uses the S&M rankings. You could play AA against four opponents who play hands from groups one through four, for example. It's called Poker Calculator and a google search should bring it up.

[pokerbase]

Good hands aren't really that much better against AA than random hands. For instance, T8o does about as well as KK.

[ninthyaga]

Good point, but I was actually just using AA as an example. I was wondering any way to improve this type of computer analysis in general.

[rigoletto]

Try: real hands database

[droidboy]

You could use Turbo Texas hold'em: http://www.wilsonsoftware.com. The Poker Calculator" http://koti.mbnet.fi/jraevaar/pokercalculator/, and the Poker Stove: http://www.pokerstove.com both have limited ability to do more flexible simulations.