Variance

[ggbman]

Aba,

Taking yourself and CTS were notable examples of excellent players, two of the best, and citing it as an example of how little variance affects your results is silly, and i don't understand how someone as intelligent as yourself is trying to use this as argument against what i am saying. You guys both play with larger edges than most players. But like i have said numerous times. dat indicates two things:

1 BB/100 is a very solid earn in high stakes limit games online these days

1 BB/100 winners can have 1000+ BB downswings without tilting at all

So agruing with me and saying that you think people whine too much is going to change that. Do a lot players attirubte their results to variance too quickly? Yes, which is why i said that you made a good point there. But just because you and a few of your very sucessful friends haven't encountered that kind of variance doesn't mean that others can't, and that it occurs legitimatley without them tilting off tons of money. You should realize that arguing with the numers i provided is silly. It's not meant to provide an excuse for people, but at the same time the most sucessful players are the most likely to have NOT encountered this type of variance logically. It doesn't mean you, CTS, ozzy and Co are not hands down the best players, but it is obviously less likely that you would have encountered as much negative variance as many others have. I don't really care why "lucky" anyone is, but i get frustrated when a great player such as yourself makes comments about variance which are just that far off, it can and does affect some people to a much greater extent than you seem ready to accept.

And BTW

[ QUOTE ]
but lost 15-20 more 100K+ pots over the course of the year

[/ QUOTE ]

[ QUOTE ]

This is actually much more improbable than you are making it out to be assuming the money went in as a 50/50.

[/ QUOTE ]

It's not more probably than i was suggesting because my whole point was that it wouldn't take much for that to [censored] up someones year. What if it was 30-40 pots? That's probably not even 2 standard deviations away from normal for people playing 500k-1 millions hands a year and it will shave 4 million dollars off their earn. We aren't arguing here.

Anyway i don't really feel like arguing about this any more, it's just stastical facts that i am citing here and i', not going to waste my time or your arguing the exent to which people exaggerate the role of variance in their downswings. We can agree that people do it to various extents, and just remember that there is no way you have seen anywhere near antying in a half million hands, no one has.

Gabe

[mike l.]

you both seem to be ignoring the fact that variance is less of a problem for a great nl player. nl is different.

[aba20]

[ QUOTE ]
aba how many hands do you have over the past year?

[/ QUOTE ]

I would guess ~750K, but I don't know for sure because I got a new computer in may and I corrupted my old database when trying to move it to my new computer.

[DeathDonkey]

Mike,

NL is only different because in most games a great NL player can have a much greater edge than a great limit player can. In the super high stakes NL games with all decent players, the great NL player's edge can drop to one more similar to a limit player's and then everything Gabe is saying is exactly the same for both players.

-DeathDonkey

[sweetjazz]

[ QUOTE ]
Mike,

NL is only different because in most games a great NL player can have a much greater edge than a great limit player can. In the super high stakes NL games with all decent players, the great NL player's edge can drop to one more similar to a limit player's and then everything Gabe is saying is exactly the same for both players.

-DeathDonkey

[/ QUOTE ]

DD, there are two features about NL that make the associated variance different from limit. First, the unlimited betting tends to give the better player at a table a bigger edge. Thus a player's winrate at various tables would be more sensitive to who is playing at the table in NL than in limit. This works to the advantage of the very best players (who would keep bigger edges in NL than limit) and works against the very good players who insist on playing in bad games where the very best are sitting. In other words, it should be easier for a very good but not world-class limit player to minimize how much he loses to Schneids than it should be for a similarly talented NL player to minimize his losses to aba, and that's entirely due to the structure of the game being played. To this extent, aba's points have some more merit in NL than they do in limit games.

OTOH, the pots in NL get much bigger in relation to the blinds than in limit. If an optimal strategy requires taking/forcing a lot of coinflips (and I don't know enough about NL to speak with much, or even any, authority here), then the variance is going to be brutal. As the game evolves and toughens up, the edges may become small enough that luck is a major factor in who succeeds and who does not.

My basic intuition with regard to no limit is that in short stack games played close to optimally, the variance will be huge because aggressively jamming will be the best strategy. In bigger stack games, the best strategy is more likely to combine aggressive pot building and stealing with trapping, and I don't know how this compares with the possibility of the pots getting really big. (Note: All of the above is likely to be severely flawed thinking.)

As far as limit, the fact that large downswings can occur once your winrate drops to the 1 BB/100 or so area or below is mathematical fact. Indeed, in the limiting case where your winrate is exactly 0 BB/100, you would experience arbitrarily large downswings (and similarly arbitrarily large upswings). No matter how far down you were, if you kept playing long enough you'd eventually get back to even; conversely no matter far up you got, if you kept playing long enoguh you'd eventually drop below zero. This is the result of simple mathematical analysis of a random walk.

[ggbman]

DD,

That is pretty much what i am saying, with the adendum that i have no idea what an attainable WR is very high stakes NL games so i cannot comment on how much variance they would have. But in limit i know what a realistic earn is and what the correspond potential variance could be.

[obi---one]

[ QUOTE ]
I think people use variance to rationalize why they are only small winners or losing. I always take the mentality that my results are based on my play. People act like there is some outside force beyond their control. Well there isn't your results are based off of your actions and nothing else. People need to take responsibility for there losses and realize they are to blame for losing and not some statistical anomaly. When I am losing or on a downswing I don't blame variance or bad luck I look to myself, because I know that I am probably off my game. I take responsibility for the losses and look to improve my game. I know this is kind of a rant but my one point that I think can help 95% of players on this board is this; Take responsibility for your play and results. Both are a direct cause from your actions.

[/ QUOTE ]

I agree with this, and think more people should look at their results this way. BUT, limit holdem is a different animal than your main games.

[DpR]

[ QUOTE ]
"1) my understanding is that the likelihood of having a downswing of X BB increases exponentially as your winrate decreases. Can any math guys confirm or deny this?"

playing with the numbers on the website ggbman linked (http://www.reviewpokerrooms.com/poke...uirements.html) it seems that it is close enough to linear for reasonable winrates. heres the numbers i plugged in.

bb/100..roll
0.25....1941
0.5.....970
1.......485
1.5.....323
2.......243
3.......162
4.......121

i used 5% risk of ruin.

[/ QUOTE ]\


You guys are funny. How is this not EXACTLY linear?

(% increase in WR = % increase in BR) Not using % is pretty silly, obviously BR goes to infinity as WR goes to zero.

EDIT: I am only referring to the numbers posted, and not representing that they are factual for poker in the real world. I do not believe that variance is necessarily fixed as WR changes.

[cartman]

[ QUOTE ]

You guys are funny. How is this not EXACTLY linear?


[/ QUOTE ]

You cant just put % on one of the axes. Linear means that for each 1 unit increase in one axis, there is a fixed amount of increase or decrease on the other axis. (As an aside, this comparison should probably be win rate vs risk of ruin instead of win rate vs required bankroll). For this example, for the relationship to be linear the additional amount of bankroll required for a decrease in win rate from 2.0 to 1.5, for example, should be the equal to the additional amount of bankroll required for a decrease in win rate from 1.5 to 1.0. That isn't the case here.

Cartman

[daryn]

[ QUOTE ]

Also, you use the word "variance" to describe how a small winner rationalizes why he isn't a big winner but I think more frequently variance causes a small winner or a slight loser to BE a big winner over some amount of hands.


[/ QUOTE ]

great point that nobody seems to want to admit