HU Cash Bankroll Question ! NLCASH

[jay_shark]

I think you do a lousy job at being concise with what you're trying to convey . You cannot prove your conjecture using words and try to convince everyone that it's right .

In fact , all I pointed out is that it's not exactly obvious that NL heads up cash is lower variance than NL sng's . This depends on many factors that you failed to bring up .

I've made an argument that it's easy to compare your variance when you're at most 75bb's deep in a cash game . It's not entirely true that when you buyin for deep , that your variance will remain lower. This was the entire premise of my argument against yours and nothing else .

[TNixon]

[ QUOTE ]
You cannot prove your conjecture using words and try to convince everyone that it's right .

[/ QUOTE ]
While this may be true, when you and "he who shall go unnamed" tried to prove anything with math (and through that thread, both of you were arguing that HUSNGs are *lower* variance than HUCASH), you both completely flummoxed the math by changing the problem entirely, and by basically doing the equivalent of trying to compare inches directly to centimeters, saying 3 centimeters is greater than 2 inches, by comparing the numbers directly without doing any sort of conversion first.

And to top it all off, after I finally spent the time to lay out the math (which really was equivalent to 100>10, duh, and *should* have been plainly obvious), you acted like I had somehow changed my point of view, and that's what you were trying to say all along, when you were *clearly* trying to make a case for cash being higher variance. (do you really need me to go back and start quoting?)

Furthermore, this statement:

[ QUOTE ]
I've made an argument that it's easy to compare your variance when you're at most 75bb's deep in a cash game . It's not entirely true that when you buyin for deep , that your variance will remain lower.

[/ QUOTE ]

Shows that you *still* don't understand all the issues on the topic, despite your smugness, because you're still trying to directly compare big blinds, without taking into account that not all big blinds are equal, and that if you want to measure variance in any way that's actually MEANINGFUL, you have to do it in dollars, not big blinds.

If playing a 10 big blind stack worth $100 is higher variance than playing a 100 big blind stack worth $100, then it would seem highly likely that a 75BB stack would be higher variance than a 100BB stack (and you can't "buy in" any deeper than that in cash), and that the play during a $100 tournaments where you start with 75 big blinds would therefore likely be higher variance than a $100 buyins at .5/1, even at the very beginning of the tournament.

[ QUOTE ]
This was the entire premise of my argument against yours and nothing else .

[/ QUOTE ]
Yes, and every time I pointed out that your premise was faulty because you didn't seem to grasp the concept that a $10 big blind cannot be compared directly to a $1 big blind, you acted like I was being an ignorant retard. And again, when I laid out math showing that what I was saying was true (even though it was all so brain-deadeningly simple that it *should* have been obvious), you acted like that's what you were saying all along.

But hey. Lets not reopen that wound. I don't want to get banned from the forum.

[jay_shark]

I got an idea that can settle this debate once and for all .

Play at least 30 cash game sessions using a full buyin (100*bb) and then compute your variance which really is individual specific . This means that your variance may very well be different than mine . Moreover , in an sng , if you know your win rate , then your variance will be fixed which is not the case in a cash game (you probably know this ) .

Before you report with your findings , you need to assert yourself that your opponent brings in the same amount as you . You must also be willing to replenish your bankroll when you lose a pot and you're opponent must be willing to do the same .

[TNixon]

This is why I was trying to get data from other people. I don't have enough of a bankroll right now to play cash (FT doesn't have $25 or $50 buy-in cash), and I've heard the rake is pretty tough to beat at $100NL, so I likely won't play that either.

[Al_Money]

[ QUOTE ]
can you please explain?

[/ QUOTE ]

You are in more marginal situations, more often.

[jakeduke]

tnixon, jay,

I don't know if this will help your discussion any, but my SD for my last 50k hands (about 40K @ 100NL and 10k@ 200NL) is 59.1ptbb/100 according to PT.

[TNixon]

ok jay, since more than happy to admit I'm out of my element here (and I'm lazy besides):

Lets see some std dev calcs for various winrates at HUSNGs.

Obviously a sample size of 1 isn't ideal, but it's better than nothing. [img]/images/graemlins/smile.gif[/img]

[LordMushroom2]

[ QUOTE ]
in cash, variance increases with the number of entrants.

[/ QUOTE ]

Judging by your other posts I figure you meant "decreases". Just pointing this out to possibly confused readers.

[ QUOTE ]
Lets see some std dev calcs for various winrates at HUSNGs.

[/ QUOTE ]

I would be interested in hearing if the math I did in a previous thread trying to find the standard deviation of 2 HUSnGs was correct or not (at the end of my 2nd post in that thread):
The thread

[TNixon]

Ok, here's what I've come up with for variance on HUSNGs:

w = winrate (0 - 1)
W = $won when you win ($105 for $110s)
L = $lost when you lose ($115 for $110s)

mean = Ww - L(1-w)

variance = (w(W - m))^2 + ((1-w)(-L - m))^2

At $110s:

For a winrate of 55%, the mean is 6, and the variance is:
(.55(105-6))^2 + (.45(-115-6))^2
5929.605

For a winrate of 65%, the mean is 28, and the variance is:
(.65(105-28))^2 + (.35(-115-28))^2
5010.005

So, the std dev for a 55% winner is $77/game, and for a 65% winner is $70.78/game

To compare this directly to jakeduke's figure of 59.1bb/100 hands, we need to do some conversion. The easiest conversion is to convert to $/hand.

59.1bb/100 hands = .591bb/hand. Ignoring his $1/$2 games (80% of them were 0.5/1 anyway), bb = $1, leading to a std deviation of $.59 per hand.

Over 802 turbos, I averaged 34.89 hands per game, and this number was pretty consistent over all levels from $22 up through $220.

Assuming this is a fairly standard number of average hands per game (which is admittedly a big assumption, but we have to use *some* number), and using the figures for a 65% winner:

$70.70/game * 1 game/34.89 hands = $2.03 per hand.

So even a 65% winner at $110s will have a std deviation of about three and a half times jakeduke's std deviation for $0.5/$1 cash over 40k hands.

Even if jakeduke plays an extremely low variance style, it seems pretty likely that HUCash is *way* lower variance than HUSNGs for the same buyin.

Much more than I would have guessed. Of course, we have exactly TWO real-world samples here, which is far from ideal. My average hand count could be much lower than most people's, and jakeduke's variance could be lower than most people's.

Again, this is why I hoped to get real stats from a bigger selection of people, but nobody responded.

[jakeduke]

actually 1bb = $2 the way PT describes it. I just realized that PT gives the $ amount - for my 100NL play the SD is $1.10/hand.