HM1/HM2 Icm '$EV Diff' wrongly imported?

[zamb3zi]

Turbo or Hyper Turbo?

Most EV fixes tackled bugs with calculating EV for hyper turbo's.
Though whether you play Turbo or Hyper Turbo, the latest HoldemManager build has the most accurate EV.
(there still are some bugs.... only affecting splitpots... and Omaha)

So the $695 is the accurate EV Diff. And $1044 was bugged.

Almost entirely Turbo with perhaps 10 or so hypers mixed in. I would have been aware of anything too weird happening in those.

Ok, so I wouldn't have seen these EV bugfixes until I finally re-imported all my hands? Wouldn't it be an idea for the software to correct previous imports (automatically or via some db maintenance option) when you fix bugs in this area? It would have been good to know sometime in the last 6 months that I could access these bugfixes by re-importing.

[Patvs]

You'll always have to reimport your hands into a new empty database after such a EV bug fix.

HM1 used to have a 'database updater', but they were tricky to program, and had the risk of a database corruption.

[kampiuceris]

Once again I don't know how you guys work, but it's over the year ev calculations are not fixed.
This is another common bug when players have less than big blind and goes all-in (Appears the same in Hm1 and Hm2):

Playing Fifty50 ICM = { 0.6, 0.1, 0.1, 0.1, 0.1 }

PokerStars Hand #113236530453: Tournament #880342361, $57.25+$2.75 USD Hold'em No Limit - Level XI (150/300) - 2014/03/14 2:50:28 EET [2014/03/13 20:50:28 ET]
Table '880342361 1' 10-max Seat #4 is the button
Seat 3: bilaboy86 (556 in chips) 
Seat 4: Skod (4157 in chips) 
Seat 5: kampiuceris (2621 in chips) 
Seat 7: STEVAN AA (5531 in chips) 
Seat 9: valodya87x (105 in chips) 
Seat 10: Sunny220v (2030 in chips) 
bilaboy86: posts the ante 40
Skod: posts the ante 40
kampiuceris: posts the ante 40
STEVAN AA: posts the ante 40
valodya87x: posts the ante 40
Sunny220v: posts the ante 40
kampiuceris: posts small blind 150
STEVAN AA: posts big blind 300
*** HOLE CARDS ***
Dealt to kampiuceris [Jd Kc]
valodya87x: calls 65 and is all-in
Sunny220v: folds 
bilaboy86: folds 
Skod: folds 
kampiuceris: raises 2281 to 2581 and is all-in
STEVAN AA: folds 
Uncalled bet (2281) returned to kampiuceris
*** FLOP *** [Qs 6d Td]
*** TURN *** [Qs 6d Td] [6c]
*** RIVER *** [Qs 6d Td 6c] [4s]
*** SHOW DOWN ***
kampiuceris: shows [Jd Kc] (a pair of Sixes)
kampiuceris collected 470 from side pot
valodya87x: shows [9h Ad] (a pair of Sixes - Ace kicker)
valodya87x collected 435 from main pot
*** SUMMARY ***
Total pot 905 Main pot 435. Side pot 470. | Rake 0 
Board [Qs 6d Td 6c 4s]
Seat 3: bilaboy86 folded before Flop (didn't bet)
Seat 4: Skod (button) folded before Flop (didn't bet)
Seat 5: kampiuceris (small blind) showed [Jd Kc] and won (470) with a pair of Sixes
Seat 7: STEVAN AA (big blind) folded before Flop
Seat 9: valodya87x showed [9h Ad] and won (435) with a pair of Sixes
Seat 10: Sunny220v folded before Flop (didn't bet)

I can repeat once again: hm1 always calculates things correctly when pressing that "ICM" button, never saw it being wrong:



Additional correct calculations if needed:

[PreFlop] Board[] Dead[] Pot=435
kampiuceris          Jd Kc      41.60%     180.98    
valodya87x           9h Ad      58.40%     254.02    

[PreFlop] Board[] Dead[9hAd] Pot=470
kampiuceris          Jd Kc      100.00%    470       

Total chips in all pots: 905
Payouts: { 0.6 0.1 0.1 0.1 0.1 }
Buy In: $60.00 ($57.25+$2.75)
Total Prize Pool: $572.50


PlayerName        Pocket       Chips Diff      Chips Won       Chips Ev        Chips Ev Diff   Icm Ev Diff $         
---------------------------------------------------------------------------------------------------------------------
bilaboy86                      -40             0               0               0               +0.0170750  $+9.78     
Skod                           -40             0               0               0               +0.0004099  $+0.23     
kampiuceris       Jd Kc 41.6%  130             470             650.98          180.98          +0.0071335  $+4.08     (Hero)
STEVAN AA                      -340            0               0               0               +0.0002162  $+0.12     
valodya87x        9h Ad 58.4%  330             435             254.02          -180.98         -0.0270205  $-15.47    
Sunny220v                      -40             0               0               0               +0.0021860  $+1.25     

##################### Regular Ev #####################
##################### Preflop (Evaluations 1712304) #####################

Player Name         Icm Ending   ($)            
bilaboy86           0.0931636361 ($53.34)       
Skod                0.2366573460 ($135.49)      
kampiuceris         0.1961882281 ($112.32)      
STEVAN AA           0.2727296527 ($156.14)      
valodya87x          0.0380004691 ($21.76)       
Sunny220v           0.1632606680 ($93.47)       

-------------------------------------------------------------

JdKc  9hAd      probability 58.17711% (996169/1712304)
1     0     
Player Name         Stack       Icm                 
bilaboy86           516         0.0760886858        
Skod                4117        0.2362474032        
kampiuceris         2751        0.1890547610        
STEVAN AA           5191        0.2725134991        
valodya87x          435         0.0650209852        
Sunny220v           1990        0.1610746657        


JdKc  9hAd      probability 41.38599% (708654/1712304)
0     1     
Player Name         Stack       Icm                 
bilaboy86           516         0.1172000000        
Skod                4117        0.2372333333        
kampiuceris         3186        0.2062000000        
STEVAN AA           5191        0.2730333333        
valodya87x          0           0.0000000000        
Sunny220v           1990        0.1663333333        


JdKc  9hAd      probability 0.43690% (7481/1712304)
0     0     
Player Name         Stack       Icm                 
bilaboy86           516         0.0899649693        
Skod                4117        0.2366836209        
kampiuceris         2968.5      0.1976929054        
STEVAN AA           5191        0.2727458282        
valodya87x          217.5       0.0396291206        
Sunny220v           1990        0.1632835556        




Regular EV = 
(0.58177 * 0.18905476) +
(0.41386 * 0.20620000) +
(0.00437 * 0.19769291) 
= 0.196188228061929 ($112.32)
Hero after hand = 0.189054761027256 ($108.23)
Regular EV Diff = 0.196188228061929 - 0.189054761027256 = 0.0071334670346728 ($+4.08)

There is another known bug which comes to my head now: when more than 4 plays goes all-in, ev is not calculated at all. So please, please fix these things, they are indeed very easy to do. If you need coding help I can support code parts from my own implementation.

[Patvs]

Do you also have a handhistory example of a 4-way allin?

[kampiuceris]

Do you also have a handhistory example of a 4-way allin?

I said "when more than 4 plays goes all-in" what means 5-ways or more. Here is an example of 5-ways:
Playing Fifty50 ICM = { 0.6, 0.1, 0.1, 0.1, 0.1 }

PokerStars Hand #105000587930: Tournament #797185286, $57.25+$2.75 USD Hold'em No Limit - Level IX (100/200) - 2013/10/04 6:37:47 EET [2013/10/03 23:37:47 ET]
Table '797185286 1' 10-max Seat #4 is the button
Seat 1: superking168 (2090 in chips) 
Seat 2: ykkös karhu (2745 in chips) 
Seat 4: Xodd (3145 in chips) 
Seat 5: memmec (2890 in chips) 
Seat 7: kampiuceris (3180 in chips) 
Seat 8: EvgenyKo (950 in chips) 
superking168: posts the ante 25
ykkös karhu: posts the ante 25
Xodd: posts the ante 25
memmec: posts the ante 25
kampiuceris: posts the ante 25
EvgenyKo: posts the ante 25
memmec: posts small blind 100
kampiuceris: posts big blind 200
*** HOLE CARDS ***
Dealt to kampiuceris [Ad Qs]
EvgenyKo: raises 600 to 800
superking168: calls 800
ykkös karhu: calls 800
Xodd: folds 
memmec: calls 700
kampiuceris: raises 2355 to 3155 and is all-in
EvgenyKo: calls 125 and is all-in
superking168: calls 1265 and is all-in
ykkös karhu: calls 1920 and is all-in
memmec: calls 2065 and is all-in
Uncalled bet (290) returned to kampiuceris
*** FLOP *** [3h 7h 6c]
*** TURN *** [3h 7h 6c] [Qh]
*** RIVER *** [3h 7h 6c Qh] [5d]
*** SHOW DOWN ***
memmec: shows [Kh Kc] (a pair of Kings)
kampiuceris: shows [Ad Qs] (a pair of Queens)
memmec collected 290 from side pot-3 
ykkös karhu: shows [9h 9d] (a pair of Nines)
memmec collected 1965 from side pot-2 
superking168: shows [Tc As] (high card Ace)
memmec collected 4560 from side pot-1 
EvgenyKo: shows [Js Jc] (a pair of Jacks)
memmec collected 4775 from main pot
ykkös karhu finished the tournament in 4th place and received $57.25.
superking168 finished the tournament in 5th place and received $57.25.
EvgenyKo finished the tournament in 6th place
memmec wins the tournament and receives $278.43 - congratulations!
Xodd finished the tournament in 2nd place and received $116.79.
kampiuceris finished the tournament in 3rd place and received $62.78.
*** SUMMARY ***
Total pot 11590 Main pot 4775. Side pot-1 4560. Side pot-2 1965. Side pot-3 290. | Rake 0 
Board [3h 7h 6c Qh 5d]
Seat 1: superking168 showed [Tc As] and lost with high card Ace
Seat 2: ykkös karhu showed [9h 9d] and lost with a pair of Nines
Seat 4: Xodd (button) folded before Flop (didn't bet)
Seat 5: memmec (small blind) showed [Kh Kc] and won (11590) with a pair of Kings
Seat 7: kampiuceris (big blind) showed [Ad Qs] and lost with a pair of Queens
Seat 8: EvgenyKo showed [Js Jc] and lost with a pair of Jacks

hm1:



hm2:



The thing is that in this case even magical hm1 ICM button doesn't work:



The true calculations of this hand: pastebin.com/gbDY9civ
Both hm1 and hm2 calculates ev diff for $0.00, the true value is: +$32.66
For the bug note my hm1 and hm2 databases both were made (not backup-restored via pgadmin, but freshly imported by HH and TS .txt files (what means EVs were recalculated)) around 2 months ago. I was always up to date on hm1/hm2 versions, so you could guess the version if it makes any difference, for instance if there were changes in ev calculations between last 2 months.

[Patvs]

I wrote up a separate bugticket on these issues, HM-8091.

[kampiuceris]

Jesus Christ... guys? How long does it take to fix these easy bugs. It's more than year and a half. In case 'bugticket' (if they at least are usable..) got missing:

Payout: Fifty50 (60 10 10 10 10)

PokerStars Hand #122915785024: Tournament #997911076, $57.25+$2.75 USD Hold'em No Limit - Level XI (150/300) - 2014/10/10 9:18:53 EET [2014/10/10 2:18:53 ET]
Table '997911076 1' 10-max Seat #1 is the button
Seat 1: gallochapin (170 in chips)
Seat 3: mattei13 (3925 in chips)
Seat 5: subin7728 (1205 in chips)
Seat 6: Alex.Z(A3) (9415 in chips)
Seat 8: MrChallenger (15 in chips)
Seat 9: kampiuceris (270 in chips)
gallochapin: posts the ante 40
mattei13: posts the ante 40
subin7728: posts the ante 40
Alex.Z(A3): posts the ante 40
MrChallenger: posts the ante 15 and is all-in
kampiuceris: posts the ante 40
mattei13: posts small blind 150
subin7728: posts big blind 300
*** HOLE CARDS ***
Dealt to kampiuceris [Js 3c]
Alex.Z(A3): folds
kampiuceris: folds
gallochapin: folds
mattei13: raises 300 to 600
subin7728: folds
Uncalled bet (300) returned to mattei13
*** FLOP *** [2d Ks 2s]
*** TURN *** [2d Ks 2s] [Kc]
*** RIVER *** [2d Ks 2s Kc] [7d]
*** SHOW DOWN ***
mattei13: shows [5s 9d] (two pair, Kings and Deuces)
mattei13 collected 725 from side pot
MrChallenger: shows [Qc 2c] (a full house, Deuces full of Kings)
MrChallenger collected 90 from main pot
*** SUMMARY ***
Total pot 815 Main pot 90. Side pot 725. | Rake 0
Board [2d Ks 2s Kc 7d]
Seat 1: gallochapin (button) folded before Flop (didn't bet)
Seat 3: mattei13 (small blind) showed [5s 9d] and won (725) with two pair, Kings and Deuces
Seat 5: subin7728 (big blind) folded before Flop
Seat 6: Alex.Z(A3) folded before Flop (didn't bet)
Seat 8: MrChallenger showed [Qc 2c] and won (90) with a full house, Deuces full of Kings
Seat 9: kampiuceris folded before Flop (didn't bet)

HM1/HM2 shows: -$0.01, where the real answer is: +$3.39 (you can check the math here: CloudEv )
By the way magical HM1 replay "ICM" button works in this case with the correct answer.

[Patvs]

HM-8091 is still an open ticket.

Unfortunately it's not seen as a 'critical' bug, so I have no eta when this will be fixed.