Why does my equity increase here?

[MaltbyStu]

Hopefully this is the right forum.

I have QQ and 1 opponent whom I put on JJ+,AK+.

According to stove my equity is 47/52. Clear enough.

Now, I have opponent number two who will call with 65% of hands and the original caller.

Why does my equity increase to make me favorite to win at 45/38/15?

Thanks,

Stu

[jay_shark]

You mean player two calls with the top 65% of all hands ?

I'm getting that q's wins about 39.57% and that jj+ ak wins about 44.406% . The third player wins about 16.024%

[MaltbyStu]

[ QUOTE ]
You mean player two calls with the top 65% of all hands ?

I'm getting that q's wins about 39.57% and that jj+ ak wins about 44.406% . The third player wins about 16.024%

[/ QUOTE ]


This is whjat I was trying to explain.

QQ v JJ+,AK+ I am behind...

equity win tie pots won pots tied
Hand 0: 47.366% 45.79% 01.58% 164644764 5675268.00 { QQ }
Hand 1: 52.634% 51.06% 01.58% 183588540 5675268.00 { JJ+, AKs, AKo }


Add in the 65% drooler and I'm now ahead. what do I misunderstand....



equity win tie pots won pots tied
Hand 0: 45.625% 45.78% 00.17% 176020710 657148.17 { QQ }
Hand 1: 39.630% 39.64% 00.27% 152432848 1031374.67 { JJ+, AKs, AKo }
Hand 2: 14.744% 14.54% 00.31% 55912872 1183052.67 { 22+, A2s+, K2s+, Q2s+, J2s+, T2s+, 94s+, 84s+, 74s+, 64s+, 54s, A2o+, K2o+, Q4o+, J6o+, T7o+, 97o+, 86o+, 76o }

[BradleyT]

You stopped your calculation too soon.


Text results appended to pokerstove.txt

197,026,696,944 games 268.831 secs 732,901,700 games/sec

Board:
Dead:

equity win tie pots won pots tied
Hand 0: 39.943% 38.76% 01.18% 76370271948 2327679214.00 { QQ }
Hand 1: 44.136% 42.84% 01.30% 84401287032 2557862764.00 { JJ+, AKs, AKo }
Hand 2: 15.921% 15.53% 00.40% 30589744620 779851366.00 { 22+, A2s+, K2s+, Q2s+, J2s+, T2s+, 94s+, 84s+, 74s+, 64s+, 54s, A2o+, K2o+, Q4o+, J6o+, T7o+, 97o+, 86o+, 76o }

[MaltbyStu]

This kept me up all night. I just ran it again and waited patiently this time.

Thanks for checking this.

Stu

[jtr]

This is very weird. Is it evidence of a flaw in pokerstove's algorithm?

It seems a very large change in the outcome given that the initial calculation was already over 320 million or so deals. That's a pretty big sample size, for the outcome to change on the larger sample by about 6% in EV. Surely this only makes sense if pokerstove was sampling hands in some systematic way in the smaller sample?

[Siegmund]

Depends what you mean by "increasing."

Your chance of winning went down from 47% to 45%. Your financial results improved from 2x.47=0.94 units to 3x.45=1.35 units. Your first opponent's chance of winning went down from 52% to 38% (but he too showed a financial improvement, from 1.04 to 1.14).

Most of your wins with QQ come when your opponent has AK and fails to flop a pair, or when your opponent has JJ and neither of you improve.

The third player, who is usually playing crap, is winning by making a big hand - making a set with a small pair, or making a straight or flush with small connectors or Axs.

Your 2% loss comes from those rare times when you had QQ vs AK on a board with no ace or king, but that set/straight/flush pushed you down from best to second-best hand. The times when you were winning heads up because nobody improved, you are still winning three-handed.

There are more times -- QQ vs AK where an ace or king came on the flop, QQ vs AA, and QQ vs KK, where you were in second place already, and wiseguy's lucky hand moves you down from second to third.

Nothing wrong with pokerstove, just a situation where you are hurt less than your opponent is by the addition of a third player.

EDIT: I see it's been re-run, and the numbers have shifted somewhat.
I have left my original post above, because it is still possible for one hand to suffer less from the addition of a third player than another. The percentages are not so extreme as 2% loss vs 14% loss after the recalculation was done.

[DarkMagus]

For preflop evaluations with 3 or more players, always use monte carlo, because "evaluate all" evaluates hundreds of billions of cases, and it evaluates them in a systematic order and so you can't trust the results if you stop it early. Monte Carlo evaluates it randomly and after like 10 million hands the results are going to be accurate to around 0.01% or something.

Evaluate all is usually fairly quick for post-flop evaluations, and pre-flop evaluations involving only 2 players.

[jtr]

[ QUOTE ]
For preflop evaluations with 3 or more players, always use monte carlo, because "evaluate all" evaluates hundreds of billions of cases, and it evaluates them in a systematic order...

[/ QUOTE ]

Thanks a lot, that answers my question.