EQ calc in multiway pots is still false (example inside)

[Andimann]

I have played this hand this morning and was surprised about the EQ as i was all in ;)

http://sophia.zintzsch.de/Poker/EQErrorHM.gif

and in hand replayer (stove says the same - 59% )

http://sophia.zintzsch.de/Poker/EQErrorHM_Replayer.gif

All hole cards are known (at the end, not as i was all in at flop). But for the EQ calculation should be all needed information there.

I tried option "fix ev for current player" in HM but nothing changes.

I assume the calculation is done at turn as all players are all in - but thats wrong for me because i am all in at flop and there i was a 60% fav and not a 4,5% underdog.

Is there any possibility to change/correct this?

Thanks
Andreas

[netsrak]

The calculations are made when nobody is able to make any more actions and 2 or more people are all-in.
No way to change this for a hand.

[Kaitsu]

Formula for EV calculation, as it is now in the software, gives false long period result. Therefore it is totally useless for tracking players own performance. It can be only used to examine EV in single pots that are played in certain fashion (both players all-in at the same time). In omaha there are often more than 2 players in big all-in pots and long time result are more often skewed than not.

There has been several topics reporting the shortcomings in this feature. Yet I have not seen one post where you have agreed to correct this (sorry if Im wrong). In my opinion calculating sklansky bucks street by street is the best way to determine players own performance and such feature would be very important. Skansky bucks calculation would not have to be real time, if it slows the software down too much.

[netsrak]

I'll forward this to the developer.
I know there have been lots of discussions about this.

[Rvg72]

I have played this hand this morning and was surprised about the EQ as i was all in ;)

http://sophia.zintzsch.de/Poker/EQErrorHM.gif

and in hand replayer (stove says the same - 59% )

http://sophia.zintzsch.de/Poker/EQErrorHM_Replayer.gif

All hole cards are known (at the end, not as i was all in at flop). But for the EQ calculation should be all needed information there.

I tried option "fix ev for current player" in HM but nothing changes.

I assume the calculation is done at turn as all players are all in - but thats wrong for me because i am all in at flop and there i was a 60% fav and not a 4,5% underdog.

Is there any possibility to change/correct this?

Thanks
Andreas

There are two other players still active when you went all-in so calculating your all-in from that period would be invalid and provide biased results. and calculating EV% from there would be completely misleading.

Roy

[Rvg72]

Formula for EV calculation, as it is now in the software, gives false long period result. Therefore it is totally useless for tracking players own performance.

This isn't true, in fact it is completely accurate in what it tries to do which is evaluate luck once all the cards are known and players actions are complete. Some people argue that going street by street is better but more people argued the opposite when we were building this and we agreed with them. The way PokerEV did this 2 years ago (using street by street) was flawed - even the author of the program finally admitted that this was the case.

We may provide an option to go street by street in an update but saying the current method is flawed is completely incorrect.

Roy

[Kaitsu]

You have done more work on this so I trust your judgement. Still how is the current method compared better, compared to the situation where EV is calculated back for each player from the situation where one has gone all-in.

Example: P1 stack 200 (JJ), P2 200 (TT), P3 100 (QQ)

P1 raises, P2 calls, P3 ai, call, call

Flop AJ5: P1 ai, P2 fold.

P3 has roughly only 2 outs so his EV at this point is negative with current calculation. When in my mind P3's EV should be calculated from the point where he went all-in and therefore be positive. P2's money in the pot is dead money.

If all players go to showdown comparison is made between players in pairs at the point where final action for each player had been made. P1 vs P2, P2 vs P3 and P1 vs P3 separately end combined EV for each comes from there.

I probably still haven't grasped all this, but please humor me.

[Django]

Hi!

I totally agree with Kaitsu on this issue. This is really annoying bug and does not give real picture of your game play. Especially in Omaha many player situations are so common that all-in EV does not follow real action. Improvement this quality would be really huge step in analyzing own game play with HEM.

[Rvg72]

You have done more work on this so I trust your judgement. Still how is the current method compared better, compared to the situation where EV is calculated back for each player from the situation where one has gone all-in.

Example: P1 stack 200 (JJ), P2 200 (TT), P3 100 (QQ)

P1 raises, P2 calls, P3 ai, call, call

Flop AJ5: P1 ai, P2 fold.

P3 has roughly only 2 outs so his EV at this point is negative with current calculation. When in my mind P3's EV should be calculated from the point where he went all-in and therefore be positive. P2's money in the pot is dead money.

If all players go to showdown comparison is made between players in pairs at the point where final action for each player had been made. P1 vs P2, P2 vs P3 and P1 vs P3 separately end combined EV for each comes from there.

I probably still haven't grasped all this, but please humor me.

The main reason we don't do it from preflop in this spot is we don't know what P2 had. What if he had KK and folded on this A high flop - is it then fair to calculate equity from preflop and base it on QQ vs JJ?

EV always generates a lot of debate and no implementation of EV calculations from any program has not had a lot of complaints against the validity of it. The only other program that had a working EV calc (ie even doing something as basic as factoring in side pots) is the old PEV program that worked with PT2 and their method, AKAIK, would not calculate your hand from preflop here either because of the unknown player. What they did do is calculate street by street equity which has some fans and some detractors saying it introduces too much bias.

We will provide more options for people with regards to EV in the future because it is something that everyone has a different opinion about.

Roy

[Kaitsu]

The main reason we don't do it from preflop in this spot is we don't know what P2 had. What if he had KK and folded on this A high flop - is it then fair to calculate equity from preflop and base it on QQ vs JJ?


Of course it is fair because in that case KK folded and his EV = 0. Calculation should be done with known factors and known facts. In this case QQ vs JJ all-in on the flop. I think this is not nearly as big a problem in HE than it is in omaha.

Anyway I believe we're on the same page here and I'm glad to hear something is going to be done to improve the calculation. Keep up the good work guys!

[DedMazay]

OK, get same problem - I went all-in preflop with 27% equity, 2 other players went all-in for side pot on flop where my equity slipped down to 6%. HM calculated my EV based on 6% equity, not on 27%. Horribly wrong IMHO.

[strannikspb]

There are two other players still active when you went all-in so calculating your all-in from that period would be invalid and provide biased results. and calculating EV% from there would be completely misleading.

Why would it be invalid?
1. We know their cards.
2. We're all-in on the flop.
What else do we need for calculate our EV? It's no matter for us what they do on the other streets. Their actions do not influence our EV.
And what we really would like to have - is our all-in EV when WE goes all-in, when we MAKE A DECISION. And we want to know the EV of our decision.

[Olaf]

http://sophia.zintzsch.de/Poker/EQErrorHM.gif



http://sophia.zintzsch.de/Poker/EQErrorHM_Replayer.gif



Khm... We made our decission - Allin, when we had 59% to win.
We saw ALL opponents cards (nobody fold) on Showdown and still know that we had 59 - Not 4% for win against them in AllIn moment.

EV $ Diff 3.47 seems like a joke.... - We must won 59 times from 100 in this situation....

I don't know, how we must make right calculation when anybody fold on next streets after our allin, but in this situation EV $ Diff 3.47 is obvious error.

[strannikspb]

I couldn't understand anything.
Just a minute ago I went all-in on the preflop with AJo. 2 guys called. One pushed on the flop with pocket 66. The other folded on the flop. HM says that I have 92,6% Equity. Because I catched trips on the flop. But I went all-in on the preflop. And it's coinflip vs 66. And also we do not know the holecards of another opponent, that folded on the flop. So in this situation we can't calculate our EV. But HM can - 92,6%. Very strange.

[yngwiescruggs]

Of course it is fair because in that case KK folded and his EV = 0. Calculation should be done with known factors and known facts. In this case QQ vs JJ all-in on the flop. I think this is not nearly as big a problem in HE than it is in omaha.

Anyway I believe we're on the same page here and I'm glad to hear something is going to be done to improve the calculation. Keep up the good work guys!

Mostly agree with the above, but either way there is absolutely no reason we can't calculate EV on a hand where all the hole cards ARE known. Right now HEM says your EV is zero if you get in a 3 way pot, 2 players who have you covered call, and then they check it down.

The QQ vs JJ example I also tend to agree with. Over the long run, usually it's not going to be that the KK will fold here often. It's more likely you might end up with KK vs QQ, and your EV estimate will be really close. While it's not a perfect calculation, I think it's definitely better than saying your chance of winning the hand is zero.