Was there a claim about higher winrate implying a lower variance?

The coinflip vs a a 100 sided die argiment is not convincing.

Lower variance isnt brought about by betting all even numbers + 99 in the 100 sided die, but sitting plenty of them out and betting all even + primes when the opportunity presents itself.

I can agree that a higher winrate for vpip hands should imply a lower variance.

I might have missed the point of the variance in the context of it affecting your game. Without a true idea of winrate, does it even matter?

As for AIEV, stay away from it. The only true estimate of your winrate is your winrate.

Forgive me for being a little slow *cue Dumb & Dumber clip* but your contention is that using effective AIEV calculations allows for more bias in your AIEV line that simple @AI EV calc method? I get the point on how it can be manipulated and am not thinking at a high enough/thorough enough level to think it through multiple iterations, but it makes sense.

Also to clarify the variance aspect, you're referring to range variance in villain holdings correct?

No.

I also have wondered this thing a long time.. I see no difference from where the money comes as long as it comes.

Also checked Nikachu for the first time. I love that fella.

e: a slightly ot, but it's good for any thread

both are bull**** maine... But the winnings line at least is real. Real= true

true ev is the metafactor. ITs cool to try and imagaine a true ev line

if no cards were dealt it woul dbe your edge relative to a villain pool, and then only factored by position, in that it would be a perfect slope but with 6 differences cycling for each position.

when you know the opponents then it is the same but the gradients would be different and each new table would offer new gradients,

then you add tilt/non static games ect, ability to adapt ect and now the true eve line is very very hard to imagine correctly

Once the cards area dealt to each position, true ev becomes infinitely complex..

if you simplify it to a heads up game against yourself, then I guess it would be like 55:45 45:55 55:45 45:55 each hand, but once the cards are dealt, AIEV and winnings will be sooooooo far away from this.... it will actually appear random and any sample would have a very great chance of showing that one player has an edge, when he doesn't.

Over a finite sample luck is a meta factor. Perhaps even over an infinte sample it is possible for luck. 1st person observence cannnot see true ev, only the truth, and that is the past.

Just wanted to briefly comment on the redline/winrate discussion:

When I started to show marginal profit at PLO100, my graph looked like a xmas tree or a pyramid with SD and non-SD heading in opposite directions. The advice I got at the time were mainly about increasing aggression and playing less exploitably. Well, what happened was a significant increase in both my winrate and red line. Co-incidence? Hardly.

The so-called nits will always win at poker, but there's a limit to how much they can win. If you only concentrate on pots where you have guaranteed equity to win at show-down, you're leaving money on the table.

Statistics are used for analysing your game and AIEV is statistically more accurate. Using your real winrate for anything instead of AIEV (except seeing how much exactly you've won) is really dumb.



There is no bias in the method AIEV is calculated. You can't play a certain style to cheat AIEV. There is variance involved it, random luck or whatever you want to call it.

SbS EV has systematical error, making it completely useless.


For example if a player with 10bb/100 edge played 10 million hands, his actual winrate with certain confidence rate could be within 10+-0.5bb/100, his AIEV 10+-0.3bb/100 and his SbS EV would be 14+-0.1b/100. Now you probabaly see why it shouldn't be used?

GGARJ:
You are right, it's not completely useless and I didn't say that. It's simply one of the most overrated stats in my opinion. For example in this thread a 6bb/100 graph with positive redline is posted and everyone is going crazy about it. A little later 14bb/100 graph is posted with negative nonsd and asking what he is doing wrong, and nobody cares (or even bothers to mention that the winrate is over double)

Sorry chinz gonna make this a bit harder for you

There's one neat way to get bias in aiev. Last year there was one psychological reseach paper on seeing the future.

http://www.newscientist.com/article/...he-future.html

http://www.dbem.ws/FeelingFuture.pdf

According to that extroverts have a better feel for the future than introverts. Assuming your regular reg is more introvert than your average fish, the fish with plays based on feelings might have a tiny edge on gutfeelings whether he's gonna hit the flush or not for example

(I never checked if anyone found the mistake on that paper or replicated the results. Most obvious way to get those results is faulty statistical analysis)

Winrate should be the only measurement how good you are. In the long run lallallaa..

There is one real (albeit very theoretical) possibility of bias in AIEV. If you played against 5 players' colluding ring, they would be able to fold certain draws when they are more likely to miss, and when they actually get that "30%" draw in, it could be because they know it's actually 40% thanks to card removal info they got from colluding.

I'm not mathematically savvy enough to figure out how big of a difference that would really make, but I've tried to think about it I can't really see it making more than like 0.01% difference in the long run, unless you are constantly playing in tables where multiple players collude against you. (and actually are competent enough to collude effectively)

Interesting calculations - and im not sure what ur conclusion is ?

im gonna post my stats later - but im having a HIGH winning redline and bigtime loosing blueline.

But im tilting a lot - im making MANY MANY -AIEV plays. When im on tilt or making moves - so im very very often making -AIEV plays and it will show when i get "caught".

But + 100k hands now im "running" over EV - im still a 6.45bb/100 winner at PLO 100 at 100 k hands. Im defo. sure that no one has figured out correctly whats wrong with the AIEV anD EV in HEM In PLO but even galfond has said its pretty meaningless. When i tilt i tilt big time - 3 times lost yearly profit at higher stakes and the ev line just looks like **** because of the tilt.

So i know im good at my stake - a lot of winning regs has sit out vs me on stars . THe only time i under EV is when im loosing - pretty weird ha ?

Im laggy and 100 percent sure that some kind of playing style + the stakes that we play at is making EV inaccurate at PLO. NLHE its pretty accurate.

When i reach 250K hands and still being a big winner , will people say you are a on a life time heater? Or that we are getting in to something with this inaccurate EV at PLO..

There is nothing wrong with AIEV. What Galfond said doesn't contradict with that. There's nothing wrong with the way HEM calculates your turn 4bet frequency either, but that doesn't mean turn 4betting frequency is an important stat.

(Also afaik the old way HEM calculated EV was wrong in some situations, but that has been fixed a few years ago. I think PT still calculates it wrong, but I'm not sure as I don't use it myself)



100k hands is almost completely meaningless sample size, and 250k hands isn't anywhere accurate either. If you want any sort of real accuracy for winrate you need millions of hands. I really suggest you try running a few variance simulations, since you clearly have no idea whatsoever what you're talking about.



Here's 10 runs of 250k hand simulation for ya:


Does it still seem to you like 250k hands is enough to offset the variance? The worst run is about +50bi while the best run is almost +350bi... and that's just from 10 runs. Make it a few hundred runs and you'll start seeing some really crazy ****.

Well then i dont understand why many says to me that im good at PLO.
Then it must be a life time heater i guess.

And i ran almost 7K under EV when i took shot at PLO 500 (most of my years winnings) so i defo. know how variance is, feels and affects the game.

Most of the time i loose is pretty much my own fault - when im playing c-d game..

If you play enough street by street poker, and is a good handreader i defo. think its possible to decrease variance. If we get it all in on the flop all the time - then yes i agree variance could be sick , but playing a lot of turns and river - getting ur opp. to fold his hand should help a lot.

If it makes you feel any better about the amount of variance in PLO, at least all 10 runs were winning. And those are made using moderate winrate 5bb/100 and maybe slightly higher than avarage standard deviation (it's mine, but I also play HU which increases it)

When playing eg. PLO100 6max with 50% rakeback you could add about 6bb/100 to that graph and slightly decrease standard deviation.

Including 6bb/100 in rakeback to the graph and using more typical 6max SD, the graph doesn't seem quite as awful anymore:

It's not really what you said, rather what you didn't say

Noone prob cared to answer the 14bb dude who was clearly just showing off . Seriously though, I'm guessing the graph didn't look like a xmas tree even though nonSD was negative, he probably makes excellent v-bets and uses position to control potsize and is not playing a strict fit-or-fold game letting marginal spots go.

Slightly losing redline is no cause for concern, but a steep downhill def is.

Winrate of 14bb/100 is sertainly nothing to be proud of when playing under plo400.

Lol what a morronic post.. Forgiven if it was ironic .

How many players will u find as 14bb/100 winners at plo 100 and 200 at +200K hands .. Please find some nicks for me so i can learn.

You clearly haven't read enough of Slimy's fat gf posts to recognize his style

Judged by ur post - i can conclude that Slimy's post was ironic..

i would be happy to have 5bb/100 at PLO and 200 after +200K hands - more than enough for me

i think a good area to focus on where suffering a 'downswing' in PLO is non showdown winnings. looks for spots to bluff/force better hands to fold - esp on the river. given that hand ranges are fairly obvious at the lower levels once the river has been dealt there are so many more profitable bluffing spots on the river. obviously don't over do it and pick the right opponents, but look to increase those non showdown winnings. otherwise you are truly reducing the game to an allin orgy fest and will be constantly flipping and this will lead to severe variance

i am trying to work on this right now as i get tired of getting my money in good and getting drawn out on and what skill is there to just getting it in good? If that's your only goal in PLO then you better have a massive bankroll and tons of patience and a understanding girlfriend

I've been wondering, would it make SlimyF's posts more or less funny if his gf actually is fat? Is she?

Don't have panorama camera so I can't confirm it.

I dont understand ur last sentence : If that's your only goal in PLO then you better have a massive bankroll and tons of patience and a understanding girlfriend .

Are you levelling or writing something serious. When u are suffering from a downswing u should look for more bluffing spots/increasing the redline.

All in on flop orgy fest can be brutal.

Has anyone with a massive database ever done a study to see whether the people running above EV actually equal people running below EV?

And yes, of course I know that almost everyone is biased towards posting the below EV graphs in public, which makes it look like every reg "runs bad". That goes without saying.

But it still doesn't mean there isn't a small chance that such a study could uncover anomalies - after all, we already know from numerous scandals that the games aren't always 100% honest.