PDA

View Full Version : The red EV line in hyper-turbo heads up sng's



TryingHard
09-22-2012, 08:42 AM
Hey, I've been just trying to understand how the EV line for HU hyperturbos on stars works.

First I thought it sums up all the moves you make - every fold, every raise fold, every all in. I thought it assigns these moves a $ EV value and than sums up all these values and finally sketches it on the graph.

But as it shows up, the red EV line only counts the all in situations and ignores the folding spots.

But the $EV which I first thought is the red EV line can be found in : Tourney --> Resutls--->Data view section---> $EV won.

So why doesn't holdem manager sketch this $EV value on the graph? It doesn't make sense to me...

Patvs
09-23-2012, 12:13 AM
To see the line in the graph turn on 'luck adjusted winnings' .

HoldemManager uses All-in EV, so only hands resulting in a showdown, with a EV Diff will be used for the $EV Won.

TryingHard
09-23-2012, 03:39 AM
But this doesn't make sense to me. Why does holdem manager ignore the folding spots? Example. Hyperturbo sng 6,85 $ + 15c ---> 500 chips each player. Player one raises 60 and I re-raise all in ---> he folds. This has an $ expected value of 60 x 0,0137$(value of one chip) = +0,822 $ EV.

Why doesn't holdem manager go and sum up these folding spots and count them up with the equity in the all-in flops and than sketch that on the graph? The value which I am referring to can be found in Tourney-Results-Data View- $EV Won.

But instead, holdem manager only counts the all in spots, and ignores these fold, which doesn't make sense to me.

Any ideas about this? I wonder if it is calculated like this in the cash game as well?

Thanks a lot!

Patvs
09-23-2012, 02:11 PM
The "EV" line shows the EV Diff + Winnings.... it's not possible to only show a EV Diff line.

HoldemManager can't calculate EV when there is no showdown. It needs to know what the holecards were in order to calculate EV. When you get an opponent to fold, he may fold the best hand.

So it's unfair to give you +0,822 $ EV when the opponent folds.
It also results in biased results, if you mix street-by-street-EV with All-in EV.


It's calculated in the cashgames like this too. Only All-in EV is calculated.

TryingHard
09-25-2012, 01:58 PM
Yeah but in heads up sng's, the EV would be perfectly accurate. Isn't there any special update to change the EV line for heads ups?

Patvs
09-25-2012, 06:29 PM
Change it how?

In all STTs (including HU SNGs), HoldemManager also takes the ICM into account. Read: http://forums.holdemmanager.com/general-support/334792-question-about-red-line.html

TryingHard
09-26-2012, 07:01 AM
Yeah, but ICM doesn't apply for heads ups, so it's just making the EV line more inaccurate. I think that's a bit sad.

Patvs
09-26-2012, 07:55 PM
ICM also applies for HU SNGs.
When two stacks go allin, where one player can eliminate the other (but not vice versa):
Player 1: 1500 chips
Player 2: 500 chips

TryingHard
09-27-2012, 04:45 AM
Sory for my older posts, I am a bit nervous that I cant logically come to a conclusion for the EV line.

As for the ICM independent chip model, as far as I was concerned it is used to express a different value of every single chip in different phases of a sng. But in heads up sng, you play cash game basicly. Every chip has constant value, even if Player one has 1 chip left and player 2 999 chips left.

http://forumserver.twoplustwo.com/58/heads-up-nl/icm-hu-tourneys-doesnt-apply-amirite-57245/

I am a bit confused...how accurate is this EV line for heads up sngs??

TryingHard
09-27-2012, 01:50 PM
And post n.2

I was just recently trying out PT3, and I see a bit of a difference in my lifetime graph for heads up hyperturbos. There are maybe 100 more tournaments on the PT3 graph, but it doesn't change the graph that much...

On my HM1 graph I am at 0 buy ins in expected value. And on the PT3 graph I am on around + 15 000 chips in expected value, where heads up hyper-turbos buy in equals 500 chips, so I think that should be 30 Buy-ins up right? I am adding both screenshots.

Also, could someone tell me where precisely the mistakes are made, so that I can check it and continue working with my EV line. Because like this I am a bit confused. Again, I play only heads up hyper-turbos, so there are no complicated calculations.

84792

84802

Patvs
09-27-2012, 01:56 PM
Cashgame - Preflop Allin
Player 1: $1500 AA (80% fav.)
Player 2: $500 22 (20 % fav.)
Player 2 wins the hand

Player 1's EV: +$400 (80% of $500)
Player 2's EV: +$100 (20% of $500)
Player 1's EV Diff: +$900 ($400 - -$500)
Player 2's EV Diff: -$900 ($100 - $1000

Buyin is $10 HU SNG - Preflop Allin
Player 1: 1500 chips AA (80% fav.)
Player 2: 500 chips 22 (20 % fav.)
Player 2 wins the hand

Player 1's cEV: +400
Player 2's cEV: +100
Player 1's EV: +$18 (80% * $20 + 20% * $10)
Player 2's EV: +$2 (80% * $0 + 20% * $10)
Player 1's EV Diff: +$8 ($18 - $10)
Player 2's EV Diff: -$8 ($2 - $10)


So EV in the cashgame is the same as cEV in the HU SNG.
But the EV and EV Diff work differently. The cashgames resolve around the effective stacksize and a $1000 total pot, player A and player B can both win $500. The HU SNG resolves around the difference between the EV and the outcome of the hand--> = the EV Diff. Player 1 can win $20, Player 2 cannot.

When I stated HoldemManager uses ICM I meant, it makes this: (80% * $20 + 20% * $10) vs (80% * $0 + 20% * $10) type calculation which uses two different equities, based on if the player can win, or can only double his stack.

Patvs
09-27-2012, 01:59 PM
The cEV line in PT3 is an entirely different line based on 15.793 hands.
Try changing the graph to number of games played.

TryingHard
09-27-2012, 04:06 PM
Patvs, thanks a lot for your answers.

As I am rethinking the whole thing, I am starting to realize that the EV line is pretty accurate for heads up sng's. Do you agree with me?

And as for the bug with hypers, should I be worried?

Patvs
09-27-2012, 08:15 PM
I made a mistake in my previous calculation.. it's (80% * $20 + 20% * $10) not (80% * $20 + 20% * $0), though it was only an example how a HU SNG is different as a HU Cashgame hand.



EV line should be accurate for all HU SNG... but we currently have an EV bug affecting hyper turbo's. Do you ONLY play Hyper Turbo HU, or also regular or TURBO HU SNGs?
If you click on the ICM button if you replay a hand, it will show the accurate numbers.

But the reports (and the graph which users the numbers from the reports) are slightly off for each hand. They're only off by a couple of cents (for each allin hand), but that can have quite an effect if you have multiple all-in hands in a game**, and have played thousands of games.

It should be fixed in the next HM1 (and HM2) update.

**the effect should be smaller in HU SNGs, and more noticeable in regular 6max and 9max hyper turbo's STTs.

huckleberry
09-28-2012, 02:46 AM
Cashgame - Preflop Allin
Player 1: $1500 AA (80% fav.)
Player 2: $500 22 (20 % fav.)
Player 2 wins the hand

Player 1's EV: +$400 (80% of $500)
Player 2's EV: +$100 (20% of $500)
Player 1's EV Diff: +$900 ($400 - -$500)
Player 2's EV Diff: -$900 ($100 - $1000

Buyin is $10 HU SNG - Preflop Allin
Player 1: 1500 chips AA (80% fav.)
Player 2: 500 chips 22 (20 % fav.)
Player 2 wins the hand (winning $20)

Player 1's cEV: +400
Player 2's cEV: +100
Player 1's EV: +$18 (80% * $20 + 20% * $10)
Player 2's EV: +$2 (80% * $0 + 20% * $10)
Player 1's EV Diff: +$8 ($18 - $10)
Player 2's EV Diff: -$8 ($2 - $10)




totally wrong

Cashgame - Preflop Allin
Player 1: $1500 AA (80% fav.)
Player 2: $500 22 (20 % fav.)
Player 2 wins the hand

Player 1's EV: $800 (80% of $1000) + $1000 not involved
Player 2's EV: $200 (20% of $1000)
Player 1's EV Diff: $800 ($800 - $0)
Player 2's EV Diff: -$800 ($200 - $1000)

Buyin is $10 HU SNG - Preflop Allin
Player 1: 1500 chips AA (80% fav.)
Player 2: 500 chips 22 (20 % fav.)
Player 2 wins the hand (winning $20 ?? how?)

Player 1's cEV: 800 + 1000 not involved
Player 2's cEV: 200
Player 1's EV: $8 (80% * $10) + 10$ not involved
Player 2's EV: $2 (20% * $10)
Player 1's EV Diff: $8 ($8 - 0)
Player 2's EV Diff: -$8 ($2 - $10)




Player 1 can win $20, Player 2 cannot.

it's like player 1 in cash can win $2000, Player 2 cannot.
Player 1 already have 1500 chips, but he can win only 500 chips = 1/4 of all chips = 1/4 prize pool = $5.
total pot = 1000 chips = $10

Patvs
09-28-2012, 03:50 AM
it's like player 1 in cash can win $2000, Player 2 cannot.
Player 1 already have 1500 chips, but he can win only 500 chips = 1/4 of all chips = 1/4 prize pool = $5.
total pot = 1000 chips = $10


If it's a winner takes all HU SNG with a $10 buyin. The winner 'wins' $20.
(you could also state he has NET WINNINGS of $10 if he wins)
In the cashgame player 1 can 'end up' with $2000, but he can only win $500.
In the SNG, Player 1 can either WIN the game, or lose 500 chips and fall back to his 1000 starting stack (worth $10 in equity).

You're over-simplifying HU SNGs if you state the 1/4 of all chips = $5 is the exact same as the $500 in the cashgame, whereas the whole point of a (regular) SNG is there is a MAXIMUM in equity based on the first price and the payout structure.
If you have more chips, the equity is distributed over the remaining players. Which does not exist in cashgames.

My post was meant as an example to show @Tryinghard to approach Luck Adjusted Winnings differently.

huckleberry
09-28-2012, 05:03 AM
If it's a winner takes all HU SNG with a $10 buyin. The winner 'wins' $20.
(you could also state he has NET WINNINGS of $10 if he wins)
In the cashgame player 1 can 'end up' with $2000, but he can only win $500.
In the SNG, Player 1 can either WIN the game, or lose 500 chips and fall back to his 1000 starting stack (worth $10 in equity).


ok, player 1 in cash game can "end up" with $2000, player 2 cannot. player 1 in hu sng can "end up" with $20, player 2 cannot. what difference?


You're over-simplifying HU SNGs if you state the 1/4 of all chips = $5 is the exact same as the $500 in the cashgame, whereas the whole point of a (regular) SNG is there is a MAXIMUM in equity based on the first price and the payout structure.
If you have more chips, the equity is distributed over the remaining players. Which does not exist in cashgames.

i'm not over-simplifying. in HU sng if you will win all chips, you will take all prize pool. so 1/4 of all chips = 1/4 of $_prize_pool. so $EVdiff in HU sng = $EVdiff in cash. but it's not work for example for 6man sng (payout structure 65/35) where if you will win all chips, you will take only 65% of prize pool.

TryingHard
09-28-2012, 08:11 AM
Patvs, is this the problem you are referring to?

84892

And my previous question was more of a general question. When EV works as it should, how accurate do you consider it for heads up sng's. That's what I was asking.

Because I am messing around with and studying the EV line for a week now and I am starting to like it, but I am still not totally confident in it.

Patvs
09-28-2012, 11:32 PM
i'm not over-simplifying. in HU sng if you will win all chips, you will take all prize pool. so 1/4 of all chips = 1/4 of $_prize_pool. so $EVdiff in HU sng = $EVdiff in cash. but it's not work for example for 6man sng (payout structure 65/35) where if you will win all chips, you will take only 65% of prize pool.

For this reason, I think @Tryinghard should approach EV in cash vs STTs differently.
If the payout would be 65%/35%, Player 1 has 1999 chips. Player 2 has 1 chip.
That single chip is worth 35% of $20 = $7





Patvs, is this the problem you are referring to?



Yes.


What it's working as it should, Luck Adjusted Winnings (in any STT) is a good longterm indication how you're running.
It's accurate in calculating All-in EV. Though at the same time All-in EV is flawed (explained here: http://forums.holdemmanager.com/manager-general/24059-ev-not-all-situations.html )
So for analysing a single game it often won't be useful at all.
It will be once you've played 2000+ games.

TryingHard
09-29-2012, 08:06 AM
And do we now approximately when the update will come out?

fozzy71
09-29-2012, 10:58 PM
And do we now approximately when the update will come out?

No, we do not. There are too many variables for us to give an accurate estimate.