Stars 6 max hyper Ev

[Colin252]

Hey I was wondering where we are on the stars 6 max hyper ev calc issue?
I keep seeing everywhere that it's broken, and that's certainly true for $300 & above where result=ev. It seems like all my games $200 & below are working correctly in 1.12.11 internal. Can confirm this is working?
When will the ev for $300+ be up and running?

With hypers now such a popular format, and the variance in them being crazy due to the amount of all in collisions the ev line for hypers becomes a very VERY important tool & it would be of great value t the SNG player pool to have this sorted ASAP.

[Patvs]

EV for hypers is not accurate in HoldemManager. (affecting HM1 and HM2)
It's disabled in some builds, but even when it is shown it is not accurate.

Bugticket on this issue is HMONE-127.
This is taking long to fix, as it's also affecting HM2, and we're trying to address all EV related bugs at once for both HM1 and HM2.

[Colin252]

Ok, as a full time player I'm relying on this feature to help me define my edge & select which games I play and study so forgive me for pressing you for a little more information on this.

Please find attatched my graph for the month. My saving grace in running this bad was that my ev was at least positive even if I lose when all in more than I should. How inaccurate is the Ev feature, and what situations is causing the inaccuracies? Are you talking a few % points here and there or would it in fact be as far out as this graph is? ie +2.9% (ev) ---> -3.1% (result)

Are the ev BB/100 and all in $ ev adjusted stats in the report accurate (not in ICM, but at least as an indication I am not winning as many chips all in as I'm entitled to)?

If for example my ev here is entirely erroneous it would lead me to seriously reconsider the game format/stakes that I play for my full time income. I look forward to hearing back as this is kind of a big deal to me.
Kind regards,
Colin

[udbrky]

I'm sorry to hear you're having issues with this and I apologize for the delay. I have escalated this to our 2nd level support techs and have explained the issues you are having and they will be in touch shortly to help resolve the problem.

[SuperSized]

I believe chip EV is calculated correctly, so look into that. If you are winning more than 2bb or 3bb/100 EV, I would think that you are generally going to be a winning player at these games.

Support have indicated in the past that EV is 25% lower than it should be. So for a majority of people, EV line should rise as far as I understand it.

Sick graph bro, hang in there.

[Patvs]

EV BB/100 is accurate... so are all the chip and cEV numbers
There is an issue where the EV line is not accurate based on inaccurate EV Diff numbers. The EV Diff is accurate if you replay a hand and click on the ICM button, but it's off by a couple of cents in the reports. (and those wrong numbers are used to make up the graph)

[Colin252]

Listen thanks for getting back to me on this guys. I've run my db for this sample through PT4 (where support tells me they have no know ev issues) & the ev is almost exactly the same.
I am running massive below chip ev over the sample which easilly explains the significant difference between ev & result lines.
Given that my average stake over this sample is $130 or something will a few cents each all in really lead to that big of a difference in ev roi?
If i get all in for $120 worth of equity then $1.2 would be a 1% inaccuracy. So likely my ev roi is out by less than 0.1%? I wouldn't really consider that "broken" in terms of it's practical use to most players. I suppose if you play $7 buy ins and it is still out a few cents per all in then it's more significant.

[Patvs]

PT4 uses a different EV calculation. Read the last post of this thread: http://forums.holdemmanager.com/general-support/378431-ev-stats.html
and The Problem With All-In EV (All-In Equity) | PokerTracker (http://www.pokertracker.com/blog/2011/10/the-problem-with-all-in-ev-all-in-equity)

The hyper turbo EV Diff bug:
Although the bug only results in a EV Diff difference of $0.01 to $0.49 (cents) for each all-in hand, the long-term effects on the EV line are significant.

Every all-in hand that is called in hyper-turbo's results in a EV Diff... including in hands between two opponents you're not even involved in. As an example, let's state in every game there are 5x hands * $0.20 EV Diff difference (caused by our bug).
So for each game you play the Luck Adjusted Winnings line deviates by $1 from what it should be.

Whether you play $7 buyin or games with an average buyin of $130. If you play 5000 games. It can result in a Luck Adjusted Winnings line that is $5000 off from what it should be.



Although it looks like a nitpicky bug where the EV Diff is wrong by only $0.17. If there are four allins in one game. Your EV is suddenly off by a lot.
In the graph of the user who reported this, his EV dropped from $1995 to $1407: a 29% drop.

#pokertracker4

[Colin252]

Uhh, so other than knowing I ran a TON below chip ev, I'm still none the wiser. I assume for the difference to be so pronounced the bias is always the same way, ie the player will tend to always run slightly below ev?

[Colin252]

Actually I just looked at 10 all ins on L1 I saw today (just in which I was involved) The difference between ICm in Replayer and ICm diff showed that 9/10 of these have the (faulty) report stat shorting hero on equity v the replayer ICm diffs. This is obviously insignificant as a number but i would have expect that number to be the other way around thus creating a falsely high ev roi.
If this tiny bunch was anything to go by then it seems it's actually creating a falsely low ev roi.
Man I am confused. I don't really want to sit here and compute a thousand all in on turn or earlier hands to test this, but it seems totally counter intuitive to what I'd expect to see.
I hate life right now. That is all.

[Patvs]

Uhh, so other than knowing I ran a TON below chip ev, I'm still none the wiser. I assume for the difference to be so pronounced the bias is always the same way, ie the player will tend to always run slightly below ev?


Yes. From all examples we've seen, on average there was a 29% drop of what the EV should be.
So the graph is showing a falsely low ROI.

...which is counter intuitive with your graph. However the other examples I've analysed were all players with a positive winnings line.
I'm not sure if the bias is the other way for a negative winnings line.


You can check this by taking a small sample of allin hands. (check both games you win and lose).
The graph is showing a falsely low ROI if the EV when you click on the ICM button always is a couple of cents higher than the EV Diff shown in reports.

[Colin252]

This is a little confusing as the values will depend on if I was lucky ie win, or unlucky ie lose the hand in question.
2 actual hands
So hand I win
Replayer says Luck = $38.10
Report says $ev diff = -$38.80 ie the amount it deducts from the result/green line
So in this case Report creates the higher value lowering my redline more than the "correct" replayer ICm button by 70c

If i lose a hand
Replayer says luck $89.74
report says $ev diff = $89.42
In this case the report creates the lower value thus lowering my redline more than ICM btn in replayer by saying losing the hand cost me 32c less than replayer says.

I know this is pretty confusing, but i want to be sure my understanding of this is correct.
These hands seem very typical of what I see when i make the comparison.
Just checking I don't have it backwards becasue obv my ev is higher than my result.
These hands i've checked out seem to indicate that if anything my ev is falsely low? I am so confused.

[Patvs]

Yes...

Replayer says Luck = -$38.10 (-$38.10 right?)
Report says $ev diff = -$38.80 ie the amount it deducts from the result/green line

Replayer says luck $89.74
report says $ev diff = $89.42


...the report always is a couple of cents lower. And those number are used to make up the graph.
So the EV is falsely too low.

[Colin252]

K got it thanks

[Cloudyman]

Is this problem still ongoing? I have a similar experience that my ev is lower than my actual winnings in 6 max hypers (very small sample size of 600) - but wanna know whether it's worth me continuing playing these - because if the ev is right, I'll stop and pay something else. If it's wrong, it's worth me continuing. Many thanks.

[udbrky]

I'm sorry to hear you're having issues with this and I apologize for the delay. I have escalated this to our 2nd level support techs and have explained the issues you are having and they will be in touch shortly to help resolve the problem.

[Patvs]

EV hyper turbo is fixed since HM build 7746.
If you ever imported hands in an earlier build, you'll have to reimport them into a new empty database to see the change.



It is still an open bugticket for HM1 (and likely will never be fixed anymore for HM1)