BB/100 variations question !
 

I was wondering this and, even if my question is maybe to general, I hope to get somekind of a answer from you guys.

If I have BB/100 of 6 after 10K hands, what is the range of my actual REAL BB/100 in the long run? Variation of 6? [0,12]

How about 25K hands? 4?
How about 50K hands? 3?
How about 100K hands? 2?

I ask this question to be able to go up a limit with, in theroy, a good confidance factor of beeting the upper limit without playing hundreds thousands hands.

Thanks a lot guys.

Re: BB/100 variations question !
 

10k hands is a fairly big amount for something like this. You wanted to find out how many BB/100 you normally earn, and you tested yourself 100 times.

Unless you've been getting a lot of good luck during that run, 6 BB/100 will be very close to your "true" average. Probably within 1 or 2 at most would be my guess.

Re: BB/100 variations question !
 

I'd say that actually 10k is very little. Any upswing or downswing can change the result a lot. During 1000-2000 hands you could get so much "extra" from a good run that your winrate is inflated after 10k hands.

I would prefer to get 50k hands in before making any big assumptions. Anyway, if you do 6 BB/100, I think it is very likely that you are a winning player, but not so clear how much.

Re: BB/100 variations question !
 

Assuming NLHE since it wasn't stated.
You're up 600BB in total. It's possible, thus, you could be anywhere from deserving -400 to deserving +1600, assuming +-10BI as a sort've "biggest realisticly possible swing", with a strong bias towards to middle. That's anywhere from -4BB/100 to +16BB/100 so you'd definitely need more hands to have an accurate idea of winrate. But your certainly a very strong favourite to be a winning player.

Re: BB/100 variations question !
 

Hrm, for some reason I thought I read he was playing limit holdem. If it's NLHE then yeah, more hands are definitely needed.

Re: BB/100 variations question !
 

[ QUOTE ]
Hrm, for some reason I thought I read he was playing limit holdem. If it's NLHE then yeah, more hands are definitely needed.

[/ QUOTE ]

If I recall correctly, the variance for limit tends to be higher because you'll get the odds to call to draws much more often.

Re: BB/100 variations question !
 

10k hands is pretty insignificant.

after 50k hands you will probably know whether you are a winner or a loser at that level.

after 100k hands you will know your winrate +- about 3

after 300k hands you will know your winrate +- about 1

these are just estimates.

you can work out your confidence etc. by doing loadsa stats that someone on the forum will know

Re: BB/100 variations question !
 

Use this formula :

xbar +- z*(s.d)/sqrt(number of trails)

xbar is your sample win rate through the number of hands you're interested in .

z is the level of confidence you'd like . Typically , we'd like z=2 or about a 95% confidence level .

s.d/sqrt(number of trials) is your sample standard deviation. Each trail consists of 100 hands , so 50k hands would be equivalent to 500 trials .

Re: BB/100 variations question !
 

[ QUOTE ]
If I recall correctly, the variance for limit tends to be higher because you'll get the odds to call to draws much more often.

[/ QUOTE ]

Doesn't playing more draws mean that you're more likely to complete your hand the expected percentage of the time, reducing variance?

Don't NLHE's occasional large pots, where you can win or lose a buyin or more in one hand, lead to more variance?

I'm not sure that I'm right... if I'm wrong please enlighten me. There's only one thing that I'm sure about, which is that PLO is most luck reliant. It's crazy how often two players are correct to raise and reraise each other all in on the flop, and then a coinflip decides a buyin.

Re: BB/100 variations question !
 

[ QUOTE ]
[ QUOTE ]
If I recall correctly, the variance for limit tends to be higher because you'll get the odds to call to draws much more often.

[/ QUOTE ]

Doesn't playing more draws mean that you're more likely to complete your hand the expected percentage of the time, reducing variance?

Don't NLHE's occasional large pots, where you can win or lose a buyin or more in one hand, lead to more variance?

I'm not sure that I'm right... if I'm wrong please enlighten me. There's only one thing that I'm sure about, which is that PLO is most luck reliant. It's crazy how often two players are correct to raise and reraise each other all in on the flop, and then a coinflip decides a buyin.

[/ QUOTE ]
Clearly a NL game with blinds of 1 and 2 will have more variance than a limit game with the same blinds. The key to answering this question is to find a limit game and a NL game where experts have the same win rate, then compare the variances of those games. Of course it's very difficult say what an expert is and what one would expect to make.

Several years ago when there were a bunch of limit books out there and no one knew anything about NL, NL would have been the game with "less variance". Now that NL games are getting tougher that might not be true.

Another useful statistic is the coefficient of variation (this is the standard deviation per 100 divided by win rate per 100). You could think of it as the luck to skill ratio. It should be noted that this number is a function of the number of hands. (By that I mean the coefficient of varation per 100 does not equal the coefficient of varation per 200).

Re: BB/100 variations question !
 

I'm surprised that there were so many useless and incorrect answers.

You need to know your standard deviation, say per 100 hands. In NL, this depends more on your playing style than in limit, but typical figures are about 85 big blinds/100 in full ring and 95 big blinds/100 in 6-max. The latter is more sensitive to playing style than the former. PokerTracker reports an estimate of your standard deviation under Session Notes->More Details.

In case you meant limit hold'em, the standard deviation is usually about 15 big bets/100 in full ring, and about 17 big bets/100 shorthanded.

After n * 100 hands, the standard deviation of your win rate is (SD per 100)/sqrt(n). After 10,000 hands, n=100, so this would be (SD per 100)/10. A rough 95% confidence interval is the observed rate +- 2 standard deviations.

I'll let you decide whether your BB/100 was big blinds/100 or PTBB/100, and determine your standard deviation.

Re: BB/100 variations question !
 

[ QUOTE ]
I'm surprised that there were so many useless and incorrect answers.

After n * 100 hands, the standard deviation of your win rate is (SD per 100)/sqrt(n). After 10,000 hands, n=100, so this would be (SD per 100)/10. A rough 95% confidence interval is the observed rate +- 2 standard deviations.


[/ QUOTE ]

I'm not surprised your answer is incorrect.

Why do you lead people astray with your talk of standard deviations?

Re: BB/100 variations question !
 

[ QUOTE ]

Why do you lead people astray with your talk of standard deviations?

[/ QUOTE ]
It's in the contract.

<ul type="square">"The good Christian should beware of mathematicians, and all those who make empty prophecies. The danger already exists that the mathematicians have made a covenant with the devil to darken the spirit and to confine man in the bonds of Hell" (St. Augustine, De Genesi ad Litteram, Book II, xviii, 37). (context)[/list]Thanks for trolling.

Re: BB/100 variations question !
 

[ QUOTE ]
[ QUOTE ]

Why do you lead people astray with your talk of standard deviations?

[/ QUOTE ]
It's in the contract.

<ul type="square">"The good Christian should beware of mathematicians, and all those who make empty prophecies. The danger already exists that the mathematicians have made a covenant with the devil to darken the spirit and to confine man in the bonds of Hell" (St. Augustine, De Genesi ad Litteram, Book II, xviii, 37). (context)[/list]Thanks for trolling.

[/ QUOTE ]

Thank you for corrupting Probability Theory and increasing the risk of ruin of all these poor people!