15 posts / 0 new
Last post
jackoneill's picture
HEM - Red Line Trouble

Hi,Just had a look at my HEM database, trying to figure out why my red line keeps going down recently - it tells me I'm way above equity, which makes me feel like ... well, some fish who just can't get it in good.So I looked at a few hands and found some things that I don't understand.Example:I played this hand against a complete spewtard, who called down super light, and also used the line check/call flop, check/call turn, donk-lead river with missed draws.Blinds were 50/100, effective stacks 1395, I minraise my button with JTo, he flats with KQo. I have 35% equity and put 200 chips into the pot pre-flop.Flop comes A9J with a flush-draw, he check/calls my cbet of 200.  I have 66% equity here and put another 200 into the pot.Turn comes a 4, pot is 800 with 1095 behind.  He check/calls another 500.  Here, I have 80% equity and I put another 500 into the pot.River comes a King and he jams 595 into 1800.  Since I've seen him take this line with a lot of weaker hands, including complete air and missed draws, I call.HEM shows me $EV diff as 0, the equity column is empty.However, isn't my chip EV in this hand 400 * 35% + 400 * 66% + 1000 * 80% + 595 * 0% = 140 + 264 + 800 + 0 = 1204 chips.  Total pot was 2990 => so I had 1204 / 2990 = ~40% equity in this hand.This was a $55 turbo, so should I have a $EV of ~$44 here ? 

jackoneill's picture
Other example, from a

Other example, from a different match.Blinds 10/20, 1220 effective stacks, I'm in the BB with KK, my opponent has J2s.Pre-Flop: I 3bet him to 120, he flats.  Here, I have 84% equity and put 120 chips into the pot.Flop (592 rainbow, pot 240), I lead 180, he calls.  I have 77% equity here and put 180 chips into the pot.He hits his Jack at the turn.  That still gives me 18% equity and we get it in here.  So I put a total of 940 into the pot with 18% equity.Shouldn't I have a chip-EV here of 2 * 120 * 84% + 2 * 180 * 77% + 2 * 940 * 18% = 200 + 276 + 338= 817 chips ?HEM says cEV diff is 451 and $EV diff is $16.53.I get exactly this chip EV if I take 18.2% equity over the total pot size of 2480.So is it possible that HEM simply takes the equity at the time the hand went all-in and multiplies it with the total pot size ?That sounds wrong to me since it doesn't account for the money that went into the pot on previous streets with a higher equity.There are plenty of other hands in my database where I pot a lot of money into the pot while still ahead, my opponent hit on a later street with less than a pot-sized bet left, we got it in and HEM gives me 0 cEV diff for the hand.Over the last few sessions, I only got very few hands all-in pre-flop, but played a lot of huge pots post-flop, where we got it in on later streets.So maybe I should stop looking at HEM and All-In luck etc., but just concentrate on my game.


qattack's picture
The problem with EV as it is

The problem with EV as it is currently calculated is...variance.The reall question is not what your EV is against Villain's particular hand each game, it is what is your EV against Villain's range each game. Since there is no accurate way to measure this, EV is not as useful as it seems.Variance with regard to EV is almost as much of a problem as variance with regard to actual results. You still need tens of thousands of hands to get an accurate picture.

jackoneill's picture
Very good point. I was just a

Very good point.I was just a bit shocked when I looked at my graph, since the EV line going down does seem to indicate that I'm doing something wrong, like getting in in bad a lot.  So seeing myself some 15+ buy-ins above equity made me think "omg, does this mean I'm expected to lose 15+ buy-ins in near future?" for a moment.But you're right, it's really about an opponent's range and not his individual hand.For instance in that 952J hand - by the time I lead the turn with my KK overpair in a reraised pot, I had no idea that I was beat - and even as I got raised, I still had to give him some random 9x hands, occasional bluffs plus Jx floats that I'm still ahead against, even a slowplayed QQ could still be in his range.Normally, I don't look at my graph that often, but today I kept HEM open a kept a looking at the red line while playing.One thing which really worried me was that often - after winning a match - my equity adjusted winnings only increased very little, like 10% of a buy-in or so.  This made me think that I might be getting it in bad and suck out on my opponents.However, really I shouldn't be too results oriented - it's not about a nicely looking graph, but about playing good.


qattack's picture
Another to make is that if

Another to make is that if you are a certain amount of buy-ins over EV, this has absolutely no effect on your future earnings.If you start your poker career by being 100 buyins over EV, then in another 50,000 games, it's more likely that you will be 100 buyins above average for your total poker career.So the saying "things even out in the long run" is not exactly true. I mean they MAY, but more likely they won't. But it IS true that from the present moment forward you will on the average be exactly at EV, regardless of past results.I'm sure you know this when you stop to think about it, but many random players would refuse to believe such an outrageous statement. That's why poker is good. People don't have a clue.

fuseo's picture
As far as I know EV is

As far as I know EV is calculated as soon as the cards are shown.The problem with fish is they call down with totally incorrect implied odds and make your EV graph look bad.  If that first guy would of shoved instead of flatting you would of had 80% equity of that pot.The red line is only one small aspect of variance. There is variance in cards dealt whilst hands are not shown that contribute more to your overall luck.  So it is possible that your graph shows you above EV (lucky) , but it is very possible you actually were unlucky in those games overall when you calculate (impossible to do) every possible aspect of variance (luck).Also, I dont think you should correlate the EV direction with how well you played poker. I mean you can shove JJ at 8BB and suck out on AA and the graph will say you were lucky, but I'd say you are pretty unlucky to come up against AA. 

qattack's picture
The method jackoneill used to

The method jackoneill used to calculate EV is a somewhat more accuracte, street-by-street method. It has been considered for use by the programmer of HEM, as well as other softwares, but the HEM programmer hasn't taken the time to reprogram it.You last point was the same point that I am making...JJ is very strong against your opponent's range there, but just got unlucky against AA. This can happen an absurb amount of the time in a small sample, skewing EV results, whereas if you calculated that same JJ against your opponent's range, you would have a nice +EV.

jackoneill's picture
Well, before I looked at

Well, before I looked at this, I had the impression that HEM calculated the EV street-by-street.  Worst thing that may happen with the current approach is that you openraise AA to 1499 chips, check it down till the river where fish sucks out on you and then get the last 1 chip in dead - this should give you 80% equity over the pot of 2998.Another thing are non-showdown winnings - no idea how HEM deals with them when calculating your expected winnings.But you're right, I shouldn't worry too much about all-in EV - and especially not look at my graph while playing.


qattack's picture
Or that same AA betting all

Or that same AA betting all but 1 chip on turn and Villain calls with 22 and hits a 2. haha. That should be 95% EV, but registers as 0 EV.I think that Pat (HEM manager programmer) doesn't deem EV significant enough to worry about reprogramming it to be more accurate when he has so many other things to work on.

jackoneill's picture
Christmas results - these are

Christmas results - these are the exact same tournaments: and this is how HEM sees it: These are from both $55 and $110. This is the graph by buy-ins in HEM:


jackoneill's picture
So according to HEM, I'm $300

So according to HEM, I'm $300 above equity, that's also what the text left to the graph says.  But when I graph by buy-ins, my expected winnings are suddenly equal to my actual winnings.And PT3 gives me 400 BB expected winnings and 250 BB actual, so I'm 150 BB below equity.  If I remember correctly, BB means "Big Bet" in PT3, so these are actually 300 Big Blind.The Turbos at Stars start at 75 Big Blind, so I'm 4 buy-ins below equity, is that right ?  Since I played most of the recent matches at the $110's, that'd around $400.Does PT3 correctly compute expected winnings street-by-street or does it suffer from the same problem that HEM does ?Running below equity would actually make me feel great since that'd mean that my game is ok and I'm expected to win more money in the long-run.I didn't use PT3 much before, but started with it after watching Mersenneary's great video.  Since I recently switched to a completely new computer installation, this was also the perfect time to create a brand new database for each of HEM and PT3.What about the Non-Showdown winnings in PT3, I lost 250 BB == 500 Big Blind == 7+ buy-ins on these - it that something to worry about, or is it normal as long as the showdown-winnings are ok ?


jackoneill's picture
Another important difference

Another important difference I just realized:These huge swings that HEM shows closely reflect the swings in my bankroll - but if I look at the PT3 graph, I was never actually down in BB - and the same also happens if I graph by chips in PT3.  And my expected winnings also were never negative in the PT3 graph.Now I'm completely confused ...


fuseo's picture
The red line is a bit of a

The red line is a bit of a joke. Last game got villain down to a short stack and I shoved his limp at 5BB with JQ and he calls with 45. Then I get him to 6.3BB again and I shove QT from the button and he calls with 26. I then shove TJ at 9.2BB and he calls with A7 and HEM tells me my red line should of gone down, sure.Anyone know of a calculator the calculates true all in luck?

jackoneill's picture
Well, the problem is simply

Well, the problem is simply that these tools operate on individual hands and not on opponent's ranges.  If you shove TJ at 9.2BB, then you're ahead of your opponent's range, but HEM/PT3 can't know that.


fuseo's picture
I have no problems with it

I have no problems with it taking into account individual hands, but I think it should take into account all past all-ins. Even if it averaged the equity won over all all-ins, that would be more accurate.Really all in ev is kind of pointless, it kind of just acts like a smoothed trend line. It may be a better indicator with superturbos.