Hi again,
thanks for the answer, here's more infos and questions
I made a test with a crafted hand. I'll do more testing and I'll report.
Quote:
Originally Posted by redlotus
Ok, I'll give. I was going by what several other posters have written. Take a look in any of the "Online Poker is Rigged" threads where EV graphs are brought up and there's inevitably someone bringing this up.
Bringing what up? That PT and HEM EV calcs are wrong in particular cases? (any link appreciated for I'd like hand history/ies showing this [bad] behavior of PT/HEM).
Quote:
I suppose my choice of words was a little harsh. AFAIK, neither PT nor HeM will generate an error if the EV calculations have a possibility of being off. SharpStats, OTOH, checks for a couple of specific situations. If it finds either of these, it marks the hand with an error bit so that the EV can be calculated manually. Here are the two situations:
Hero's net win for the hand is positive, but it is less than the EV-Adjusted net win. The two situations that cause this are 1. spots where you get it in with an equity advantage, but you end up splitting the pot (eg your AA vs villain's AK and the board comes up 2345x) 2. Pots with one or more side pots.
Hero's net win for the hand is negative, but it is not as negative as the EV-Adjusted net win. If I remember my math correctly, this can only happen when there are side pots.
But what makes you think these two cases are problematic for HM/PT? For deals going to showdown, where all the holecards are known, it's possible to compute the expected value of every pot (main pot and side-pots) for all the players.
There's a tiny bias --and this is another topic-- when a player is all-in at some pre-river street and another player folds at a later street (there's a discussion entitled "EV in a side pot, folders, all cards known" on 2+2 on that subject), but the EV calculations should never be way off.
I crafted a hand with three people all-in preflop where the hero makes a net win, but less than what is expected.
AA vs AK vs J2. All players have $0.10, total main (and only) pot: $0.30.
AA's pot equity is 76% (as verified in PokerStove). This is the number that HM gives too (what HM calls the "avg allin EV%").
So hero should be making $0.22 on average, for a net win of $0.12.
However hero splitted with the AK hand and made only $0.15, for a net win of $0.05.
So hero is "running bad" by approx 7 cents (7.8 cents actually).
These are exactly the numbers reported by HEM, which shows your net win as $0.05 and the "$EV adjusted" as $0.13 ($0.05 + $0.078, which gives $0.1278, approximated by HM to $0.13).
That crafted hand history shows a case where the net win is positive, but less than the EV-adjusted net win and HM has no problem with it. I really don't see why this case would be problematic.
I think that crafted case is exactly the "situation 1" you described. And HM has no problem extracting the correct EV infos in this case.
I've got a
lot of gripes and complains about PT and HM, but this is not one of them.
Now I haven't tested with sidepots. I'll agree with you that showing a value "average all-in ev%" is meaningless in a multi-way deal with sidepots (is it pot1 ev% + pot2 ev% average? is it ev1+ev2 average? If pot one has $10 and pot two $350, I'm pretty sure this info is completely misleading). "average all-in ev%" makes sense in PokerStove, for PokerStove has no concepts of side-pots. But HM loves to display misleading information
(I won't start about all the stats calculated by taking into account actions due to a player disconnecting).
As a sidenote PokerStove is kinda misleading too: it gives an info that it calls "pot equity" when 99% of multi-way all-ins involve multiple pots. PokerStove's "pot equity %" is actually "mainpot equity %".
Anyway, it's not because HM allows you to display that misleading information when there are multiple pots that it's actually computing the "EV diff" or "$EV adjusted" wrong in multi-way pots involving side-pots.
I'm not saying it's right: I haven't tested yet.
But I have tested your first situation and HM has no problem with it.
Quote:
I will see if I have any HHs where the EV calc is off. The big problem is, though, that I only have a couple hundred cash game hands and I'm not running a version of HeM that includes EV calcs with tournament hands.
Here's a fake PokerStars hand I crafted to test your "situation 1". Three players all in preflop, they all have the same stack. No rake for it's less than $1 pot.
Code:
PokerStars Game #14100000001: Hold'em No Limit ($0.01/$0.02) - 2008/01/15 - 01:02:03 (ET)
Table 'Fake' 9-max Seat #9 is the button
Seat 1: villain1 ($0.10 in chips)
Seat 2: villain2 ($0.10 in chips)
Seat 3: hero ($0.10 in chips)
villain1: posts small blind $0.01
villain2: posts big blind $0.02
*** HOLE CARDS ***
Dealt to hero [Kc Ac]
hero: bets $0.10 and is all-in
villain1: calls $0.09 and is all-in
villain2: calls $0.08 and is all-in
*** FLOP *** [2c 3c 4d]
*** TURN *** [2c 3c 4d] [5s]
*** RIVER *** [2c 3c 4d 5s] [Th]
*** SHOW DOWN ***
villain1: shows [As Ad] (a straight, Ace to Five)
hero: shows [Kc Ac] (a straight, Ace to Five)
villain2: shows [2d Jh] (high card Jack)
villain1 collected $0.15 from pot
hero collected $0.15 from pot
*** SUMMARY ***
Total pot $0.30 | Rake $0.0
Board [2c 3c 4d 5s Th]
Seat 1: villain1 (small blind) showed [As Ad] and won ($0.15) with a straight, Ace to Five
Seat 2: villain2 (big blind) showed [2d Jh] and lost with high card Jack
Seat 3: hero (button) showed [Kc Ac] and won ($0.15) with a straight, Ace to Five
If you find a hand where HM is way off, you can post it.
I'll try to craft a case with sidepots and prove HM wrong
I'll report my findings