The green line is what's wrong then.

The red line is simply the $EV_diff value added to the green line and if the green line is wrong then obviously the red line will be too.

Juk

No, the green line is correct, $EV_diff calculation uses the wrong payout structure...

Seriously, it seems every time I open this thread now there is somebody trying to refute the HEM red line for one reason or another?

If people just forgot about the HEM "red line" and concentrated on $EV_diff instead (it's a very simple calculation; no more complex that working out the $EV of an action), they'd quickly realize that their "quest" is a futile one (excluding HEM bugs obviously) and move on.

Juk

I'm not on a quest to prove the ev diff stat wrong at all, I know it's wrong, I did the math. I want to know if there's a way to get HM to recognize that I'm playing a DoN as opposed to a 50/30/20.

Well please post the maths then.

Juk

OK, based on your earlier post:

So ignoring the chance of an obscure multiway tie and we assume 100% chance to win:

$EV_expected = $EQ(win)

$EV_actual = $EQ(win)

$EV_diff = $EV_expected - $EV_actual = $EQ(win) - $EQ(win) = 0

You can use any equity mapping function you like here (and get the payout structure as wrong as you want too!), but as you can see there is no way the $EV_diff value can be anything but zero:
and since $EV_diff is the difference between the red and green lines by definition, what you are observing is not the fault of the $EV_diff calculation or it using the wrong payout structure.

Maybe HEM has some weird bug, but the calculation and the use of the wrong payout structure is NOT to blame.

Juk

So on this hand my "EV diff" is -20.54$ (In a 10$ tournament) which we know is obviously wrong right away because it's impossible to run any more than 19.99$ over or under EV in any given hand due to the max payout being 20 dollars.

So anyways, we can use an ICM calculator (I'm using this one: http://www.icmpoker.com/Calculator.aspx) and we see that when I win the hand my equity is $20 (obviously) and that happens 19% of the time. When I lose the hand my equity is $14.1 which happens 81% of the time.

SOOOOOO my EV in the hand is .81*14.1+.19*20 = 15.22. Since I won 20 my diff should be -4.78. My actual diff, as previously mentioned, was -20.54.

Now to test the hypothesis that it's using a 50/30/20 payout we can try it again with the same calculator and the new structure.

Our equity when we lose is now 9.7 and when we win it's 35.1. So our equity in the hand is 14.5 and if we win our equity increases to 35.1. 14.5 - 35.1 =-20.6.

So the red line is wrong in DoN's. Does anybody know how to fix it? or do you still think the red line is correct?

I accept that it is, but the post you made in post #1100 was in reference to vookenmeister's post about "variance being higher" (and in turn the earlier linked article)?

Yes, it could be the case that for DoNs using cEV or even the green-line is a better predictor but that doesn't mean there is anything wrong with the maths; it's simply your equity estimation function is too wrong to be useful.

Juk

BTW, you're not the only one frustrated by this: HEM1 still thinks that Party $11 Speeds have a $10.25+$0.75 payout structure for some unknown random reason and it looks like it's never gonna get fixed thanks to HEM2

Juk

The most frustrating part is that I emailed them last night essentially explaining the problem in the same way as I did in this thread and they sent me back a response acting like I don't know what luck adjusted EV is, so now I either have to give up or thumb through my database and come up with a bunch of examples where the EV is wrong and somehow convince them that they need to fix the problem.

I can imagine that they do get a lot of daft e-mails though so can't really blame them.

One thing that might help is IIRC HEM dumps a log file somewhere of the all-in EV calculations - you could try creating a new DB with just some key HH files imported and then send them the HH files and the log explaining the problem that way.

Juk

Yeah just looked and the log is stored in "C:\Program Files\Holdem Manager\Logs\evLog.txt" (be careful opening the file if it's very big! Notepad locks up on 100MB+ files lol) and has a line starting with "Prizes" that displays the structure.

Juk

Not only do you gusy waste time staring at the stupid red line you gotta discuss it at length too

Download Notepad++ if you need to open that file. Opens large files instantly unlike notepad which might take your computer down with it lol

My log is too big even for notepad++ :/

Grunching a bit so not sure if this is exactly the same as what you're looking for in the log -- the structure used in the ICM calc via the hand replayer.

For example, here is an example where it's using 65/35 structure that should be 70/30. (Actual payouts are $21 and $9.)

very thin i hit a royal brag

I agree Tomo, the amount of time ppl spending try to disprove AIEV or talking about what your redline "likes" and "doesnt like" is pretty funny.

I'm trying to see a counselor for my red line addiction. Unfort like all addictions it's hard to give up. I am hoping that moving to HEM2 can be the methadone for my addiction. I just can't go cold turkey

And for the record I an running 1.5% below the red line (So I feel better if it's accurate). I only have 2K sample of DoNs though they are prob te lowest variance game possible

Thank you very much for posting that. Seems like quite a good blog overall, not sure if it's yours but kudos. Anyway, good explanation of the redline, and how it's calculated. Stands to reason that a viewpoint of "it's still subject to a lot of variance, but less than you actual results" seems wise.

I do still have some questions that perhaps someone could shed some light on. How is the redline effected by:

1. Hands that don't go to showdown (ie hero raises pre, cbets, and fold to a flop checkraise, hero raises and fold to a 3b preflop, hero raises and wins the pot preflop, ect) I would assume in these instances our equity obv changes, and the redline and green line should change the same amount. I guess I'm just stating a fact here, hoping no one disagrees with me. Obv EV diff is always 0 w/ no showdown.

2. Now what happens when the money goes in street by street. For instance say we open AA and get flatted. We elect to bet bet shove a board of 852K7 and get called down by 77 and lose. What happens with our equity. What if we take the same line with AA vs 77 on 752KA? Is all the equity calculated when the $ goes in? Take an extreme example. I get all but 1chip on the flop with a 9out flush draw. My opponent calls with all but 1 of his chips. He has top set. The turn he donks his remaining t1 and I've hit my flush. How/when is equity calculated?

What I've always heard is that these situation even out over time, although I'm not 100% sure they do. Different play styles could cause this. Say you bet bet shove you starting stack on most board textures in a hyper, and another player pot pots to get stacks in by turn. If equity is calculated when $ goes in (And i'm admitting I'm not sure if it is or isn't) well then the player whoe gets the money in on 2 streets figures to have a positive redline in every instance where his hand is vs an opponent hand that isn't folding at any point but improves to the best hand by river. If hero is getting money in on 3 streets, then he gets his $ in drawing dead on all these same rivers, and his redline will not go up. I'm not saying that it's a fault with the redline per say, more just a consideration of how accurate a determinate of skill it is. Our hero betting 3 streets instead of 2 to get stacks in in these spots could be taking a more +EV line that gets weaker calls from villains holdings and valuetowns them better than the hero who pot pots, but every time the villain get there on the river in essentially the same situation we get drastically different results. Thoughts?

Another spot I'm curious about, is say it is proven by and ICM calculation to be profitable to shove ATC vs a certain villain in a certain spot. Now another player may be making a mistake by only shoving 90% in this same spot. That means I'm going to have the bottom 10% of hands occasionally getting called in spots where I lose a ton of equity with very much the worst hand. If I win, my redline is going to tank b/c I had 23o, and if I lose, the redline will go down almost as much as the green line, b/c I likely had very little positive equity. Is the idea here that if it TRUELY is profitable, all the extra blinds we pick up over time from our additional frequency shoving this 10% of hands will outweigh the occasional large losses and contribute to an overall increase in the redline over a large enough sample, relative to the same hero only shoving 90%? This would stand to reason, however it would take a fantastically large sample to come to fruition. If this is not the case then obv that's a problem.

Like i said, more accurate than results but still a ton of variance is probably a fair assessment of the HEM redline. I did see some very strange things while playing HU, but I don't have the examples to back it up. Perhaps, and hopefully, it was errors. I'd do more sessions of HU and see if I can recreate what I saw but atm I'm going for SNE with 6max and don't really have the time. If I get a chance to find any spots that defy logic I'll certainly post them ITT.

tl;dr but redline improves accuracy of results irregardless of whether or not it is calculating correctly (70/30 vs 65/35 luck is still the same pretty much). It doesn't tell you how much you should have won or any of that, it just 'converges to the mean' faster than actual results. That's it. No more, no less.

Its not my blog. I hope the owner is not mad at me for posting it.

If it didn't assume that every situation you are in is average than whats the point? Every game has its own situational luck. If the bluff you get it in with 0% equity is good 2/3 times that is reflected in the other 2 times that it works. It can't have any way of knowing how much anyone's range is so it assumes each spot is an average situation. If you find a way to be more or less down in any game than is possible it should average over time in all of the games.

So in your 77 example, there should be the other 43/45 times he doesn't 2 outter you. Where you win the pot legit (with 100% equity). Of course there is variance in how much you are 2 outtered. It is another type of luck which redline is not attempting to calculate.

It is only attempting to normalize all in ev because there is no way to know if someone would call on the river without their rivered 2 outter, so it is way too difficult to calculate. It is much more useful for it to assume all situations are average. This allows you to have a more accurate estimate over a shorter number of games than simply results.

Sorry if these responses coming off rude, I am really not trying to be rude but probably shouldn't have posted in this thread to begin with.

I didn't think your responses were rude at all. I welcome informed discussion on the topic, I'm only trying to understand it better myself so I have an idea of how much weight to place on the data my redline is showing me..

I'm playing the 3$ 6max turbo games on PP, and my red line just won't go up..
I know that 650 tourneys isn't much and that the variance is high,
but would there be any chance that HEM is calculating my red line incorrectly?
My red and green line are diverting more and more, although I don't get the impression that I'm running so super hot (obviously I'm running good just not that good), or that I'm playing that bad.
I mean, just knowing to play tight when the blinds are low and decent push/fold strategy towards the bubble should be enough to get an upgowing red line at these games I would assume, yet my line stays more or less flat.

That's a big difference but 650 is still a small sample. The red line is going to be close to accurate for what it measures. My advice is to use SNG wizard to review your bustout hands, sessions, etc. or even simpler try playing with the SNG quiz that comes built into HEM1. You're obv running good but that doesn't mean you're playing good or bad. So review hands and run them thru a tool or thru peers (because it's hard to spot our own leaks). and remember with 650 tourney you can be running bad in other spots that red line doesn't measure. Just focus on improving your game. GL

I did both of those things. Also read Collin Moshman's book and watched a ton of videos from deuces cracked/cardrunners. SNG wizard agrees with the vast majority of my pushes. I probably still fail to recognize some push opertunities, but I think I should be ok for the 3$ level.
It just frustrates me that that that damn red line stays so low. But better this than the other way around I guess..

Yes indeed. I'd say don't stress. Keep studying and be happy it's not the other way around. Not sure what ROI you have but you could b running higher than achievable by an expert. GL and consider moving up when your roll allows it