all-in ev

[freanki]

Hi,

can someone explain how the all-in ev is computed? Are those just sklansky bucks?

Best regards, Frank

[morny]

Hi, EV is calculated only for all-in hands and only considers the pot size and equity when the hand went all-in. It is a measure of luck after the hands get all-in

If you get all in as 50/50 then your expected $EV is 0 because over 100 hands youll win half and hell win half and your return will be 0, if you win this hand in lets say a $200 pot your EV is +$100 because you will win 0 on average however if you lose its -$100

[freanki]

Thank you for the answer.

But then the all-in graph is no use, as long as you don't know the winnings/looses (in $) of those situations.

You can't compare the ev graph to the showdown winnings since there are all winnings wheather you've been ai or not. You can't compare it to the real winnings either because those are combined winnings of showdown and non-showdown winnings.

Otherwise please explain how to read the all-in graph.

[morny]

Its not an all in graph, Its a winnings graph which also displays all in EV. The things you mentioned would improve the the graph and things like this will be added to it in the future however its not of no use as it calculates the measure of luck after the hand is all in. Yes if we had an expected all in EV line it would better comparrisson and would be more benifical measuring luck but like i said it will be introduced at a later stage.

[freanki]

So I read the graph like this:
As long as the graph is positiv I'm playing my all-in situations good and when it gets negative I'm playing with a negative all-in ev (for that period).

[Drogbic]

Hello,

Yesterday i check allin EV in the graphs. All hands in database were shown. I made a filter. I check "Allin at the turn and earlier". Maybe it was better for seeing Allin EV comparaison ?

But... i didnt see "River allin or earlier" in the filter.. why ?

[morny]

Your equity on the river when you get all in is already decided because there is no more cards to be dealt so your equity is usually 0 or 100% so thats why there is no allin on the river filter although i guess it would be useful for other things so i can make a suggestion to have it added.

[Yikes]

If you get all in as 50/50 then your expected $EV is 0 because over 100 hands youll win half and hell win half and your return will be 0, if you win this hand in lets say a $200 pot your EV is +$100 because you will win 0 on average however if you lose its -$100

Is there anyway to tell how someone runs in these 50/50 situations? You can flip a coin 1000 times and end up with heads 700 times, it'll even out over time but some times you get the worst of it over and over, shouldn't that be factored in somewhere into holdem manager? Yes over time it should be 0, but streaks in poker sometimes are gross.

[morny]

Not exactly sure what you mean, if you flip a coin 1000 times and 700 are against you and you wagered $1 each time then youd be running bad as your expected return is $500 and youve only got $300. Of course the more hands the better we can judge the variance but in the above scenario your running bad no matter what way you look at it.

[Cernunnos]

Hey,
So at the moment the all-in ev not usuable to determine how am i running, right? I know there isn't any software which can do that, because of the cooler situations and other things, but i mean like Poker-EV. For example i have AA, raise preflop, 1 caller, bet 859r flop, bet T turn, check 9 river, villian checks too and wins the pot with 79. At this example HEM says there's 0 EV $ difference, i guess because we didn't go all-in before river, so it just ignores this hand from the EV calculations. Of course in fact this hand was +EV for me, and there's difference between the expected value and the actual winnings.

[morny]

Ive been chatting to Roy just to verify these things and it turns out i didnt understand it properly so ill explain what the All in EV stat actually is.

Say you get all in as a big favourite or even a big dog. HM calculates how much you should win on average if you had average luck i.e your the 4/1 fav then on average you should win 4 in 5 and if your a dog at 4/1 then youll win 1 in 5 with average luck. So if you do exactly that and win 4 in 5 when your 4/1 favourite then your all in EV line will be $0, if your running hot then your all in EV will be greater than $0 and if your running bad your all in EV will be less then $0. If your playing $200nl and your +$600 in all in EV your running 3 buyins above expectation. So the all in EV line compares how much you should have won in All in situations with how much you actually won and displays the difference to tell you if your running above or below expectation or in other words running hot or cold.

We also have plans in the future, (although it wont be soon and i dont have an ETA with the current workload) to introduce a new stat that will consider our equity across preflop/flop/turn and what % of our stack was in compared to the equity at the time which will produce a fairer stat in the case where we get 80%+ of our stack in on the flop as a big favourite and then All in EV is only calculated on the turn when the final 20% is gone in so if he hits his miracle card then the calculation would be very different when in fact you got the majority of your money in as a huge favourite but HM calculates it when you were a big dog on the turn.

I apologise for the confusion

[Cernunnos]

wow, this is interesting! so the ev graph doesn't indicate the equity but the ev difference, right? Here is how i see it now:
http://i43.photobucket.com/albums/e354/Cernunnos86/evgraph.jpg
correct?

[morny]

Exactly Spot on, so we can look at this line independantly of our winnings and see if were running hot or cold

[Cernunnos]

Thank you for the answers, however something is still wrong: today in a hand i went all-in preflop with KK vs AKs, final pot was $36.10. An A came on flop so i lost the hand. In HEM at "Hands" tab in "EV $ diff" column there's $23.79. But that's not the difference, that's my equity from the pot (0.66*36.10). On the graph my Winnings line correctly goes down $18.90, but the All-in EV line goes up by $4.90. If i win the hand i win $36.10 and i'm above expectation by $12.30. If i lose i lose $18.90 and i'm $42 below expectation, so the All-in EV line should go down 12.30 or up 42, if it works like mentioned above, doesn't?

[TCA]

The allin EV, as far as I understand, shows the amount you should win with average luck (not the difference in how you are performing).

However, the stat makes no sense because of the way it is calculated which I believe is pot size * win %, i.e. if you win a $1000 pot that you are an 80% favourite to win then you are clearly running $200 good on that pot.

While I believe this logic is solid preflop, postflop this makes no sense and is skewing the EV result.

An extreme example:

I have 33 in the BB, SB has AA. Blinds 5/10 with 1000 stack. He limps, I check.

Flop 332. We have 990 left to play.
He bets 989, I call.

Turn A. Check check.
River A. He bets 1 I call.

Now all in EV would show I called all in as 0% therefore I run standard.
Yet this is obviously ridiculous due to where the money went in when I was a 99.9% favourite.

- Yes this is an extreme example but clearly highlights a huge flaw in this stat.


Also it doesn't consider money already bet. When you make a bet the money isn't yours anymore, so if you call the rest off with exactly the correct odds surely EV should be 0?? Whereas the HEM would show massively -EV. This is clearly far more difficult, I think at the end of the day creating a actual meaningful stat would be much harder.


Hope I explained that ok, did it in a hurry, hopefully my logic is correct.
Hence most players will experience EV and actual winnings diverging in the long run - correct?

[morny]

First were looking into changing it to calculate based on the % of the stack in at the time and your equity at the time so in the instance above its a better representation. Also weve also noticed from Cerns thread that it wasnt calculating as it should be and were going to fix this soon but it will be a while before we have the newer stat in pace as there so much development on at the moment.

[TheZepper]

There's obviously a big misunderstanding here of the stat E(V) All-in (before the river) which measures the variance (difference) between the expected value and the actual results of an all-in before the river. By definition, there CANNOT be a variance between the expected value and the actual value of an all-in AFTER the river card is dealt - the variance is ZERO.

TCA is confusing this stat with Sklansky Bucks, which measures the sums of the expected value of a hand street by street - therefore, it includes the difference between expected value and the actual results DUE TO THE RIVER CARD.

Cernunnos is miscalculating the variance (difference) when he loses the pot. Ignoring the possibility of a split pot for a moment, he will either win the pot and end up with $36.10 or lose it and end up with $0 - so his variance is +$12.31 when he wins and -$23.79 when he loses.

Therefore, the graph of E(v) (all-ins before the river) shows one's actual winnings +/- variance (all-ins before the river) - it is a "luck" graph and is one measure that can be used to determine whether one is running good or bad over a range of hands (or time).

Sklansky Bucks are a better measure of overall variance and one's expected win-rate, and I would really like to see it added to the current graphing choices.

While I'm at it, I concur that the variance is being miscalculated in multiway pots where the player is involved in a side pot.

Finally, since this is my first HEM post, I'd be remiss is I didn't say that I think that HEM is beyond fantastic - stats geek's heaven!

[Cernunnos]

TheZepper

EV(expected value) is just EV, Sklansky bucks = EV. It doesn't matter on which street or at which action you look at it, it's calculated the same way. There's the pot, there's a bet you have to make to win the pot (or check which is like betting 0) and there's your hand which has some equity vs villian's hand. You'r EV = (times when you win * pot) - (times when you lose * bet) or = your equity * the whole pot. It doesn't matter if the bet is rest of your stack or only 2-3 big blinds. Also it doesn't matter what happened on the preivous streets, your previous bets are now the part of the current pot and nothing to do with them. Of course you can only calculate you'r exact EV when you know villian's exact hand, if you don't, there comes the tricky part, but since HEM calculates EV only for the known hole cards, that case is irrelevant now. Yes, if we look only the river there won't be difference between EV and actual winnings, since no more cards to come, our equity is 0% or 100%, as Morny already said. However if i lose a hand i lose my bet, so i end up with -$mybetammount, i end up with $0 if we split (withouth rake)

[Travys]

The correct all-in EV in multiway pots is THE bug correction I'm waiting for, even before any others minors or majors ones you may have to do.
Even if you plan to improve the EV calculation based on the %stack involved on each street, I think you should correct the multiway all-in EV first and soon, if possible...
Am I the only one desperately waiting for this correction ? Without this, the EV is simply unusable on a medium-large sample, because of lots of miscalculated pots. Please do something !

[TheZepper]

Cernunnos: My point was that Sklansky Bucks = EV BUT that the HEM graph of "All-in EV" is a graph of ACTUAL WINNINGS +/- TOTAL VARIANCE of EV (ALL-INs BEFORE THE RIVER). Two different things altogether - thats why I'd like to see HEM offer BOTH GRAPHS.

The ALL-IN EV graph is 100% a measure of one's "luck" - obviously, there is no skill involved once all the money's in and there are cards to come. On the other hand, Sklansky Bucks are clearly a more accurate measure of Var (TOTAL) than Var (EV ALL-IN BEFORE RIVER) PLUS it is one way to measure one's "skill" (expected win-rate).

Travys: I'm unclear as to what you mean by "Am I the only one desperately waiting for this correction (EV in multiway pots)". Obviously, I am, Cernunnos is, and I got here from a post by Ben_Coy in the manager suggestions forum, so he's obviously interested in getting the correction made as well. There isn't any other "correction" to be made: the second issue is the addition of a graph of Sklansky Bucks. If you're talking about priorities, I agree 100% - correct the mutiway (side) pot issue ASAP.

Also of note is that we're talking about VARIANCE of EV here (the difference in actual results from EV) - not EV itself. If you just wanted a graph of EV you'd just graph Sklansky Bucks - you'd have no need to graph your actual results at all!!!

FInally, if it were up to me, I'd like to have the ability to add my rakeback and bonuses to any one of the 3 graphs: Actual, EV (all-in before the river) or Sklansky Bucks.

Hope this clarifies what I was trying to say.

[Cernunnos]

Ok, i know what you mean, street-by-street EV calc, just your notations confused me a bit. Yes that's a nice thing to have and more accurate. Also notice that the all-in EV graph currently bugged, but as you can see Morny's answers they're already working on it so atm the graph should be ignored. Overall it's a pretty good thing that we picky bastards worry about EV graph, because it really isn't an important thing, so it tells a lot about HEM :) The thick green line tells everything what you have to know anyway!

[hmf_1]

First were looking into changing it to calculate based on the % of the stack in at the time and your equity at the time so in the instance above its a better representation. Also weve also noticed from Cerns thread that it wasnt calculating as it should be and were going to fix this soon but it will be a while before we have the newer stat in pace as there so much development on at the moment.

Hi.

What do you mean exactly when you said "it wasnt calculating as it should be". Does that mean "don't use the All-In EV Graph right now"?

Because following the Graph I'm running really really bad. But I couldn't find a hand the EV was calculated wrong. So is there a bug or not?

Thx.
Lars

[morny]

Yes as i mentioned its not calculating correctly at the moment, You can use the graph just dont tick the All in EV graph option, you can still use all the other options in there including rakeback/bonuses/showdown winnings etc but the all in EV is not accurate so if your assuming your running good or bad based on this then you may be incorrect

[hmf_1]

But like I said, I can not find a hand where the EV is calculatet wrong. Can you tell me how I can see the bug?

[morny]

Trust me i wouldnt lie if it was calculating correctly:p. Im on about the $EV Adjusted stat in reports, the EV diff one is calculating fine but the $EV Adjusted is the one you see in the graph not EV diff. Every hand is calculated wrong with $EV Adjusted so if you just add it up then youll see its not correct. If you think its correct please attach the original HH of this hand and ill run it through HM and show you the calculations

[hmf_1]

Everything looks allright here. Did I understand anything wrong?

Sorry for asking again but I'll feel better if I understand. Thanks very much!

http://img221.imageshack.us/img221/2488/exampleyd6.jpg

[morny]

You found how it is actually calculated but that is wrong. The 1st column shows how much you actually won so if you won a buyin at 200nl it would show actual winnings as $200 in a $400 pot. the EV difference shows how much from the total pot i.e $400 you would have won with average luck.

Actually to find how much your running above or below average luck you just need to invert the EV diff stat

So in your example -$317 means your running +$317 better than average luck over that sample but its the Adjusted EV one it shows in the graph so ignore this. Anyway well try and get this all fixed as soon as possible

[hmf_1]

Maybe I'm to stupid or my English is so bad ... but I still don't get it.

For my example:
I did win $72.75 but with average luck I sould have lost $244.29. If those numbers are correct I'm running $317.04 better than I should. Right?

These are the same numbers the Graph is showing me so why is this wrong?
The fat green line shows my actual winnings and the other green line shows what I've hat won with average luck. Right?

So why is this so important for me? Over the last 3 month the fat green line is at $5.000 but the other green line is at $40.000. Should I have won $40.000 with average luck or not???

Thanks for your patience and help!

[morny]

Ok you actually helped us figure this out, we thought the All in EV (Adjusted EV) should show the amount we won/lost above/below average. So in your case if you run the graph for that sample we thought it should show +$317.04 for the All in EV line as your running $317.04 better than average luck however the stat is actually calculated by taking the amount youve won above/below expectation (317.04) and adding it to your actual winnings ($75)so in your example below if you had average luck you would have won $317.04 less than +$75 which is $244

So the graph EV line is in fact accurate the problem was how we thought it was calculated. So when you look at the graph youll see how much your actual winnings would have been with average luck and compare than to your actual winning to see if your running above or below expectation for all in situations.

Really sorry for all the confusion surrounding this over the last few weeks. If you need further explanation post a graph and i can explain better

[TheZepper]

morny - there never has been a problem with the calculation or the graph EXCEPT in multiway pots. In the case where Hero is involved in a side pot, both the Equity% and the EV $ Diff figures are incorrect.

[Chamouth]

Hello - this is my first post, sorry if this question has already been addressed.

I thought I understood the calculations behind all-in EV in Hold'em Manager until I played this hand:

$0.10/$0.20 NL Hold'em Cash Game, 6 Players
LeggoPoker.com (http://www.leggopoker.com) - Hand History Converter (http://www.leggopoker.com/hh)

UTG: $14.32
MP: $28.42
CO: $28.57
Hero (BTN): $25.21
SB: $9.60
BB: $23.28

Pre-Flop: A:heart: 7:heart: dealt to Hero (BTN)
UTG folds, MP calls $0.20, CO folds, Hero calls $0.20, SB calls $0.10, BB checks

Flop: ($0.80) 7:diamond: 6:heart: 5:heart: (4 Players)
SB checks, BB checks, MP bets $0.80, Hero raises to $3.20, 2 folds, MP raises to $28.22 and is All-In, Hero calls $21.81 and is All-In

Turn: ($50.82) 3:club: (2 Players - 1 is All-In)

River: ($50.82) J:heart: (2 Players - 1 is All-In)

Results: $50.82 Pot
MP showed Q:heart 8:heart and WON $3.21 (-$22 NET)
Hero showed A:heart: 7:heart: and WON $48.28 (+$23.07 NET)


In short, Hero (me) has 65.8 % equity when all the money goes in on the flop and wins the hand. He called an all-in move from Villain for the remaining of his stack, ie. 21.81 $, with 5 $ already in the pot. Pot odds are slightly better than 4:5 with roughly 2:1 equity. Clearly a +EV situation.

My EV here is 31.99$ (when I win, ie. 65.8% of the time) - 16.62$ (when I lose, 34.2 % of the time) = 15.37$.

Hold'em Manager gives very different values though. In my sessions frame, I have 23.07 in the '$' column ($ actually won - rake), with an 'EV difference' of -16.53$.

I don't get it. Where does that number come from ? I thought it would compute EV($) - $. If so, it should be -7.7.

Can someone please explain me what happened ?

[Ben_Coy]

I think it works like this:

Let's say the raked pot is $48.28. (Just guessing from the hand-history).

Your "EV" is ($Profit X Chance of Winning) - ($Loss X Chance of Losing)

"EV" = ($48.28-$25.21)X0.658 - $25.21X0.342 = $6.56

EV $ Diff = Actual Amount Won - "EV" = ($48.28-$25.21) - $6.56 = $16.51

I believe this is the value shown in the "EV $ Diff" column. I cannot be sure since I have not seen the formula used but I'm guessing this is what it's calculating.

Actually they must be calculating it as "EV $ Diff" = "EV" - Actual Amount Won. That would account for the minus sign.
So you should see that whenever you win a hand you have negative "EV $ Diff" and when you lose it's positive.

[morny]

Yeah Ben is correct as it shows a positive result with a minus and vice versa for EV diff, if you attach the original HH ill import it and show you exactly how its calculated.

[zzz]

Does the all-in EV graph works correct on sites where you are not able to see mucked cards in hand history? For example I shoved AA preflop, someone called and I won, but I don't know his holecards, because he mucked, so actually we cannot estimate the equity. EV $ Diff in those situations is 0$ so it doesn't count in our graph, right?

[Olaf]

I had situations like this EXTREME example:

Preflop. I had AA, maniac - A4. NL 0.5$
MY stack 60$, Maniacs- 50.5$

I bet preflop, He 3Bet 50$, I think that he is AllIn and call:

Flop 444 and he bet his 0.5$... I have 0 Equity.

Sometimes situation like this happened, and EV calculation work bad for them :(

[morny]

It cant calculate equity if it dosent show the hands.

Were aware of the issue with the extreme example and will work on improving the stat to compensate equity across all streets at a later date

[TheZepper]

Morny - I believe that this thread is a perfect example of the level of confusion out here regarding variance and EV calculations. There are 3 entirely different and separate concepts involved and a lot of people obviously have a very difficult time differentiating between them.
I believe that the biggest problem involves terminology or nomenclature (although "Sklansky Bucks" is quite distintive). The "All-in EV stat" or "Diff in EV" or EV(all-in before the river) or VAR(EV All-in before the river) or "luck" stats or graphs just haven't got the job done.
Therefore, I think that right after you guys fix the multiway-pot calculation (an easy fix) but before you start working on adding Sklansky Bucks to your menu (a very ardous and time-consuming task), HEM should come up with a distintive, catchy name for the concepts relating to the "Diff in EV or all-in EV stat and graph".

Maybe something like "All-in Suckout Lucre", "Fish Funds" or "Donkey Dollars"?

[morny]

Yeah probably a good idea, in the mean time ill setup a FAQ over the next few days which explains in details what the All In Ev means and EV difference

[TheZepper]

On second thought, maybe something useful and practical like a FAQ might be a way better option.....

[corrado]

Yes, I'm a spewtard, but I have the feeling that I'm getting coolered a lot lately and that I'm running pretty bad over the last 80k hands.
Can anybody from the team can confirm that this by having a look at this graph. The thick green line would be below 0 if my non-SD-winnings were 0 but would the thin green line drop too? any help/indication appreciated.

http://img84.imageshack.us/img84/2225/wtfpa7.jpg

[fabio]

You are running 3k above expectation, sir

[TheZepper]

What the heck are your playing stats like? What game are you playing? I've never seen a graph where nonSD winnings were so high - especially in comparison with your SD winnings.

Yeah - u are running about 3k above EV for the 140,000 hands. However, I don't understand ur reference to "the last 80k hands" - did u mean 140K?

[torytrae]

http://img220.imageshack.us/img220/8148/evdifzk0.jpg (http://imageshack.us)

Ok, I think I found an error in your calculation.
It is easy to see in the KK hands where I have 80.8% EQ. I won a pot of 54.25$, but of course my net win is "only" 25,80$ as shown correctly. Now you adjust my EV $ according to the 80.8% EQ - that is fine.

But you use the pot for it, not my net winnings. So you get the 19.2% I would have lost calculated by the whole pot which is around -9.89$

Now you counter my net-winning by the 19% of the pot = +25.8 - 9,89 giving me a net-win of only 15,91 for the statistics.

This is inconsisting, either you give the whole pot as my winning and do this, or you calculate my net winnings, then only adjust EQ on my netwinnings - not on pot.

In reality my +25.8 would ony be substracted by 19.2% = 4.9$ giving me a net-win according to my EQ of 20.9$

This difference is getting more severe on the QQ hand where I have 71.6%. You substract -21$ (EQ of whole pot of 77$) from my net-winning whereas it should ony by -10,7$.

Like this you get an insane sum of -EV adjusting - thus showing on every player out there -EV $ because you double their substraction.

[morny]

Ill try and simplify this for you:

EV Diff = your equity% of the total pot when you get all in and comparing that to what you actually won

SO AK ($100 pot x 45% = $45 EV Diff)
And 55 ($42.70 x 84.1% = $35 EV Diff)

All In EV = Actual amount won - or + the amount you would have won with average luck

AK (-$50 + $45) = -$5 is what youd win with average luck
55 (-$21 + 35) = + $14 is what youd win with average luck
Total expect win on average = $14 - $5 = $9

Actual Amount = -$71
Expected Amount = +$9

So your running $80 worse than expected for someone with average luck

[torytrae]

Thx, I was making the wrong point. Please check my edit :)

[morny]

Ok ill check this with Roy for clarification

[torytrae]

Sorry for the confusion. I think I was wrong.
Its a bit difficult to understand and I just realized, that I had the wrong ideas about EV and EQ.

So lets take a standard-hand: KK vs QQ pf All-In (100$ pot)
so each player put in 50$.

The EQ of the 100$ is 80%. So it means KK expects a win of 80$ (EV $).
Now my error was assuming, that I risk 50$ having 80% EQ resulting in a net-win of 40$. For example:

I play this hand 10 times. I win 8 and lose 2.
So results would be 8x50$ = 400$ + 2 x (-50$ loss) = 300$.

300$ netwin != 800$ EQ

After thinking it over I realized:
My average pot win is 80$ in this hand (80% of 100$). So in average I risk 50$ to win 30$ net. Which is 60% of my 50$ I put into.

So finally, 80% EQ of a pot means 60% net-win of what I put in. I just understood this now -.-

Confusing for me. :eek: But I think this is correct. So probably your calculation is correct as well.

[sxe]

I have a question again, cause i didn't get it so far. :P

My problem is, i don't know how i exactly have to read the graph.

Is my luck / bad luck the difference between "Winnings" and ALL IN EV or the difference betweens "Showdown Winnings" and ALL IN EV?

If the "ALL IN EV" graph shows the EV if two players ar all in = showdown, than it seems logical to me if i have to look at the difference between "ALL IN EV" and "Showdon Winnings an not the difference between "Winnings" an "ALL IN EV" like mentioned some sites before.

Thx

[morny]

Its the difference between actual winnings and All in EV line

All In EV = Actual amount won - or + the amount you would have won with average luck so if you should have won 200 in an all in pots over a 10k sample and you actually won 300 then your all in EV line will show $100 higher than the actual amount you won. So in this example if your winnings over those $10k hands was $950 then your EV line will be $1050 ($100 higher cause you wont 100 more than you would have if you ran normal)

[sxe]

Its the difference between actual winnings and All in EV line

All In EV = Actual amount won - or + the amount you would have won with average luck so if you should have won 200 in an all in pots over a 10k sample and you actually won 300 then your all in EV line will show $100 higher than the actual amount you won. So in this example if your winnings over those $10k hands was $950 then your EV line will be $1050 ($100 higher cause you wont 100 more than you would have if you ran normal)

Thx for you answer morny. I get what you explain in the all in situation, but how dos the non showdown winnings affect the whole thing? The Winnings graph shows my non showdown winnings + my showdown winnings. But the luck faktor isn't calculated by hands where i'm not all in. When i now compare the winnings graph with the all in ev graph, whats going on with the non showdown winnings?

[morny]

The all in EV is calculated based on All in pots only and is the adjusted amount you would have won so its actual + or - your luck factor, the showdown pots has nothing to do with any of this as this also includes hands that werent all in and got to showdown.

What your missing is it take the actual amount you won and adjusts what you should have won ONLY in all in pots, then it takes the actual amount won and adds or subtracts based on the luck factor to form the all in ev line, again showdown and non showdown winnings are completely seperate.

[Demeis]

Its the difference between actual winnings and All in EV line

All In EV = Actual amount won - or + the amount you would have won with average luck so if you should have won 200 in an all in pots over a 10k sample and you actually won 300 then your all in EV line will show $100 higher than the actual amount you won.

I read this sentence about 10 times and I finally think that if you replace higher by lower, that makes sens

[guitarizt]

nm

[fraz8000]

Has the issue regarding multiway all in pots been fixed yet? This is still a common source of error for the EV calculation for players in games where shortstackers are playing, and could be easily fixed.

Even just a checkbox for 'ignore multiway allins' would be useful, and would minimise the effect of the error on statistics.


There's a bit of confusion about EV calcs in this thread. For those who don't understand how this has an effect, here's an example hand where the EV calculation is way off (ignoring rake): (Cliffnotes at bottom)


3-handed $100 NL game.
Button: $100 stack
SB: $20
BB: $100

Button has KQs,
SB has 33
BB has 88

Preflop: Button raises to $4, SB shoves for $20, BB calls and Button Calls too.

Flop (Pot $60): KQ4.

BB checks, button bets and BB folds.

Turn and River come blanks and Button wins $60 pot

Here the EV calculator attempts to estimate the equity that both players that reach showdown (SB and Button) have on the main pot. The problem is that the BB sees the flop, and then folds when they do not make their hand. Thus HEM cannot see their folded cards and so estimates the main pot equity for just the two players it can find: SB with 33 and Button with KQs.

Thus the SB is shown in pokerstove as having close to a 50% claim to the $60 pot as it is recorded as basically a 50-50 coinflip, when in actuality the SB only had about a 17% claim to the pot if the BB's hand were included. When the SB loses, the software records them as 'unlucky' for the amount of $30, when in actuality they only got unlucky to the amount of around $10.35. Also, if the river were to come a 3 and the SB wins, they would be recorded as being $30 above EV instead of $49.65 above EV.

The same happens to the button, but I won't go into it here (its just the same sort of calculation). There is obviously a bias in the EV calculation since both players who reach showdown have their EV calculated as if they were the only 2 players contesting the main pot, when in fact there were 3, and the 3rd player would have continued on had they made a better hand.

Any hands that are folded after one or more players is all in on a previous street can therefore cause inaccuracies in the EV stat in anything except for drawing-dead situations. In all cases the players that see the showdown are recorded as running worse than they actually are, whether they win or lose. Although these situations are fairly uncommon, the aggregate effect of this (particularly i games where shorstackers are common) accumulate over time, and can systematically make players believe that they are more unlucky in all-in EV than they actually are.

***Cliff's Notes***
- Any situation (except drawing dead situations) where a hand is folded after an allin from a different player on a previous street causes mistakes in the all-in EV stat.
- The effect is that all players that do see the showdown are recorded as running worse (or running less good) in all-in luck than they actually are.
- This situation is more common in games with shortstackers, and the effects can accumulate to make players think they are more unlucky than they actually are, even in the long-run.
- Problem can be fixed with an option to ignore these situations for the all-in EV calculations.
- Sorry for the long post, I just want my records to be accurate.

[morny]

Its still on the to do list, lots of implementations going on at the moment so its hard to give an accurate ETA although i will put forward your suggestion to ignore multiway pots as a temporary solution

[googly_moogly]

Morny.. Can you please help me with this example? It is important to me.

Here is the graph (if you can imagine it)

NON Showdown Winnings (red line) = +12000
Winnings (thick green line) = +9500
Showdown Winnings (thin blue line) = -3000
All-in EV (thin green line) = -4500

These are all rough numbers.. assume its a 3000 hand sample at 25/50 (if that matters)

I realize I ran really well. But what does this all mean?? On Paper, Should i have lost $4500? Did i run 14,000 above expectation? How does Non Showdown winnings factor into this graph surely it has a major effect on it? What else can this graph tell me? Assume this graph, this 3000 hand sample, goes on for infinity, with all these lines going off on the same slope as they show now.. Is this the graph of a losing player? Help me figure this out thanks!

[morny]

This is a very common question so i wrote a FAQ to explain it http://www.holdemmanager.net/faq/?f=68

[GreedyGenius]

The EV adjusted stat includes hands when there was no allin and we folded.
That makes it unusable.

Looking at my stat it shows I'm running 21.61 above expectation which is wrong. I'm running 19.61 above but it included two hands in which I folded on flop. In one I lost 0.50$ and in second 1.50$.
So it works like this.
For each hand in database add $ won to EV$ Diff and add it to EV adjusted.

http://img210.imageshack.us/img210/6885/evadjustedxd9.th.gif (http://img210.imageshack.us/my.php?image=evadjustedxd9.gif)

And I was thinking if it was possible I was that dumb and lucky that I was running above 16BI after 60k hands. This stat makes no sense at all.

[morny]

Email me the hands where it shows EV diff where you didnt get all in, certainly dosent happen for me morny@holdemmanager.net

[GreedyGenius]

EV Diff is calculated OK. It's 0 when I folded on flop. But what I put in pot preflop is subtracted from EV Adjusted.

Just look at the attached screenshot.

[Ben_Coy]

This is a very common question so i wrote a FAQ to explain it http://www.holdemmanager.net/faq/?f=68

I just read this FAQ and it really clears things up. Thanks for writing this up, anyone who is confused about the EV calculations should read it!

Also I wanted to say that I hope that future versions of HM incorporate the notes you mentioned at the end of the FAQ, i.e.:


2) If you get 95% of your stack in on the turn and the final 5% in on the river then it takes your equity on the street you got all in on, we will be introducing another stat which will try to take into consideration your equity based across all streets which should give a better representation
3) Currently Multiway Pots are not calculating properly but that will be fixed

Looking forward to seeing these improvements.

Thanks,

Ben.

[GreedyGenius]

Ok reread it 2nd time. Now I know what causes troubles. It's the name of the graph. Seeing it is called Allin EV suggests it is only from allin hands but it's not.

Why isn't it named EV Adjusted like in Reports Tab?

[morny]

It does only include all in hands:confused: I dont think you understand it properly, in that graph you won $46.13 but your equity was only 30.3% so from the pot of $92 approx you should have won 30% of that 92 on average which would be approx $28. Since you won the hand you won 92-28=$66 so you won $66 more from the pot than you would with average luck.

That explains that, now EV Adjusted which is the same stat as All in EV on the graph is your actual winnings +/- your EV difference so thats $44.13 - $65.74 = -$21.61 what thats saying is that if you played that hand 100 times or 1000 times etc on average you should win $28 from that $92 pot, since you won $92 thats 65.74 more than you should have so if you hadnt been running hot in this hand you would have a winnings of -$21.61 and not the $44 you did actually win.

[GreedyGenius]

First You say

It does only include all in hands:confused:


and then



That explains that, now EV Adjusted which is the same stat as All in EV on the graph is your actual winnings +/- your EV difference

actual winnings +/- EV Difference. But actual winnings are from all hands not only hands that went allin.

[morny]

The "it does include only all in hands" was in response to your graph you posted where you said it was including those hands, it dosent caclulate equity based on these hands but yes it does include your overall winnings

It does state this in the FAQ on the first line aswell:


All In EV Line in Graph (Also $ EV Adjusted stat in the reports TAB)
Its calculated by taking the actual amount won and either deducting or adding the amount you would have won with average luck.

Again the All in part only refers to the equity calculations and we dont do any claculations for hands that didnt get all in before or on the turn but you can also have equity calculated when the hands are not all in so we name it all in ev to differentiate all in equity to standard equity

[deletedaccount]

I'm still really confused by this.

I'm looking at a hand that has me getting it all in KKvsAK pre for 100bb stacks at 200nl and winning yet the all in EV for that hand is shown as -120.48.

How can that be possible?

Am I right in thinking it should be +140 or so in sklansky bucks? (I expect to win 70% of the time)

I just can't see 120 being acurate, mainly becasue there's no way to reach that figure in a 70/30 situation.

[morny]

Please go back through the thread, ive explained this a good few times already, please quote any one of the calculations i worked out and then write down your calculations and where you see the error, youll need to include all the details such as stack sizes, equity etc

[deletedaccount]

Please go back through the thread, ive explained this a good few times already, please quote any one of the calculations i worked out and then write down your calculations and where you see the error, youll need to include all the details such as stack sizes, equity etc

It's a really simple hand. I have 70% equity when all the money goes in pre flop. I've read the thread and I've read the ev faq.

I'll plug in my numbers into one of your examples tomorrow and double check.

[deletedaccount]

I think it works like this:

Let's say the raked pot is $48.28. (Just guessing from the hand-history).

Your "EV" is ($Profit X Chance of Winning) - ($Loss X Chance of Losing)

"EV" = ($48.28-$25.21)X0.658 - $25.21X0.342 = $6.56

EV $ Diff = Actual Amount Won - "EV" = ($48.28-$25.21) - $6.56 = $16.51

I believe this is the value shown in the "EV $ Diff" column. I cannot be sure since I have not seen the formula used but I'm guessing this is what it's calculating.

Actually they must be calculating it as "EV $ Diff" = "EV" - Actual Amount Won. That would account for the minus sign.
So you should see that whenever you win a hand you have negative "EV $ Diff" and when you lose it's positive.


Your "EV" is ($Profit X Chance of Winning) - ($Loss X Chance of Losing)

"EV" = $200 X 0.70 - $200 X 0.30 = 80

EV $ Diff = Actual Amount Won - "EV" = 200 - 80 = 120

oh.

wat.

[deletedaccount]

But I expect to win 140 with average luck, not 120. Don't I?

[morny]

With average luck you would win $280 from the $400 pot but you won $400 which is $120 more than you should have

[deletedaccount]

With average luck you would win $280 from the $400 pot but you won $400 which is $120 more than you should have

Cheers. I worked this out last night and I think it's finally sunk in.

With maths like mine it's a good job I'm running hot :D

[Blazing_Saddler]

I am struggling with how it works to be honest.

I got all in pre flop against two other Players. I had AA, they had KK and JJ respectively. It was a huge pot, think I won something like $90 at $25 NL. According to the EV on Holdem manager it was minus $17 or something along those lines. How does that work ? I checked and I had 67% equity in the pot pre flop which is when all the money went in

[morny]

Ok ill lock this up because this thread has run its course. Everything is calculating correctly but it seems a case some people are just posting "this dosent seem correct" without actually trying to calculate it and theres lots of examples in the thread that show you how to calculate as there is in the FAQ so if youve a calculation youve worked out that isnt correct start a new thread and show me the exact calculation you used to work it out and ill have a look but everything you need is located within this thread and the FAQ 10 times over