Pokerstars Double or Nothing EV Calculation Wrong

[purplechip1]

Ive been playing $5 + 0.20 Double or Nothings on Stars

There have been multiple instances of the EV won been calculated wrong.

If im correct all EV won differences on a Double or Nothing SitnGo should be between $0 and $10 as these are the min and max you can actually win in a game.

Following screenshots show differences of up to $11.35 and -$0.66.

?????

[morny]

There's no reason why the ICM luck can't be more than the buyin. The ICM EV only calculates one part of the variance in poker.

If you run like god and win every all-in coin flip, but get coolered and bust with KK vs AA, your "all-in ev luck" will be high, even though most people would say you were unlucky over all. That's because your "cooler luck" more than cancelled the "all-in ev luck".

Unfortunately "cooler luck" is impossible for us to calculate. We just give you what we can calculate. ...

[The Minder]

Originally Posted by MikeChops
There's no reason why the ICM luck can't be more than the buyin. The ICM EV only calculates one part of the variance in poker.

If you run like god and win every all-in coin flip, but get coolered and bust with KK vs AA, your "all-in ev luck" will be high, even though most people would say you were unlucky over all. That's because your "cooler luck" more than cancelled the "all-in ev luck".

Unfortunately "cooler luck" is impossible for us to calculate. We just give you what we can calculate.
I am stunned by this admission. Singlehandedly, Mike has just shot down any argument that attempts to validate EV calculations for tourneys.

[purplechip1]

im not sure you answered my question or not.

i understand that holdem manager cannot calculate EV on coolers but I dont think thats the problem here.
If im playing a $5.00 double or nothing tournament where I either cash for $10or $0 than that must be the parametres of the EV won.

But for instance I win the double or nothing, I win $10, my EV won shows that I should have won $11.

Since it is not possible for me to win $11 and my EV won is telling me I should have won more than there must be something wrong with the EV won calculations.

Am I missing something??

Or is the tournament EV in holdem manager simply unreliable?

[Patvs]

By allowing the $EV Won to be higher than the actual amount you can win/lose, the $EV Won becomes more reliable. When we introduced EV, most people were in favor of this type of calculation.

Example:
You play a Double or Nothing.
10 players. 1000 stacks. $10 buyin
First hand.. you go ALLIN with AKs.... 22 calls. (we'll call it a 50/50 coinflip to simplify the math)

If you win you have a 2000 stack.. with 9 players left (average stack 1111)--> and your equity in the DON is around $13. (the remaining $7 equity gets distributed to the other 8 players)
If you lose you lose $10.

The cEV (chip EV for the hand is 0).. so if you win, you cEV -1000. If you lose it's +1000.

The $EV difference is ((50% x $13) + (50% x $0)) = +$6.50
So if you win the hand it's -$6.50, if you lose it's +$6.50.

Next hand... another player eliminates someone else.
The hand after that.. you tangle with the other 2000 stack player in another coinflip.

If you win that hand also.... now you can end up with a $EV Won that is higher than your actual buyin... (for example Total $EV difference -$12)
Since you can end up with a combined $EV difference of -$12... the opposite must also be possible.

That usually happens.. when you lose two of more allin hands where you were a 80+% favorite, buy still manage to have chips left (because you lose against shorter stacks).... and you WIN back your chips.. in hands with nonshowdown winnings. If you now lose another hand in the bubble, where again you were a huge favorite, you can end up with a +$12 (or more) $EV Won (although you can only win $10).

Is it fair... to focus on the $EV Won for this one game? No... it's a useless , meaningless number for one game. But if you play thousands of (TURBO) SNGs, and DONs... the $EV Won for ALL the games, will be more accurate this way.

[purplechip1]

ok that makes sense
thanks