is 1200 hands shorterm?

[SplawnDarts]

[ QUOTE ]
[ QUOTE ]

Looking at the odds of your hand winning after the fact doesnt account for times that you fold winning hands, that you make other people fold winning hands, that you get free cards, that you clean up outs, that you miss value bets, make up value bets, pay off or any other meaningful measure. Most hands dont go to showdown.

[/ QUOTE ]

Again, you're largely in error. In order to re-process a hand history, you would replace results with expectations in cases where you could do so in a mathamatically exact way. No information would be lost from the history for the purposes of determining win rate. Card related variance, however, would be removed.

[Mhoram]

Isn't OP basically asking if he's just been unlucky for the last 1200 hands? And you can figure smoe of that out in the way Splawn was suggesting (if i'm understanding), atleast with Poker Tracker. You can see how many pocket pairs you had and how many turned into sets. You can see how many four flushes you had and how many turned into flushes. Same with outside straight draws. You can see how many times you had the best hand and your opponent drew out on you. All of these have a hard statistic and if you are that worried about your play you can take the time to look them over and see if any of the numbers are way off their standard.

I remember my first week of running bad I had three sets in 3000 hands. Average is 1 in 128 (if you see the flop with every pocket pair). Running short on sets is devestating in a cash game. That sort of information helps you deal with your losses and wait for the luck to break. Also should get you thinking about how you played those hands and if there was any way you could have lost less money than you did, so atleast you can improve in one area during a spell like that.

[SplawnDarts]

[ QUOTE ]
Isn't OP basically asking if he's just been unlucky for the last 1200 hands? And you can figure smoe of that out in the way Splawn was suggesting (if i'm understanding), atleast with Poker Tracker. You can see how many pocket pairs you had and how many turned into sets. You can see how many four flushes you had and how many turned into flushes. Same with outside straight draws. You can see how many times you had the best hand and your opponent drew out on you. All of these have a hard statistic and if you are that worried about your play you can take the time to look them over and see if any of the numbers are way off their standard.

I remember my first week of running bad I had three sets in 3000 hands. Average is 1 in 128 (if you see the flop with every pocket pair). Running short on sets is devestating in a cash game. That sort of information helps you deal with your losses and wait for the luck to break. Also should get you thinking about how you played those hands and if there was any way you could have lost less money than you did, so atleast you can improve in one area during a spell like that.

[/ QUOTE ]

That kind of analysis, while not specifically what I was going for, will give you an excellent idea of how the cards broke. That info, combined with observed win rate, is much more meaningful than winrate alone.

[Abbaddabba]

[ QUOTE ]
Buy a statistics book. Everything you said is completely wrong. The standard method for doing such confidence intervals is to use the OBSERVED SD, since it is the best (and only) estimator of the actual SD.


[/ QUOTE ]
Regardless of whether it's the best one at that moment in time, it isnt a GOOD estimator of the true stdv after 1k hands.

After 5 hands i have a sample stdv. And it is the "best" estimation i have of the true standard deviation. But i'd be a fool to pretend that it's a GOOD estimate. Well, it might be. But it would be completely coincidental.


It's not complicated. Im not sure why you're having trouble understanding.

[ QUOTE ]
Again, you're largely in error. In order to re-process a hand history, you would replace results with expectations in cases where you could do so in a mathamatically exact way. No information would be lost from the history for the purposes of determining win rate. Card related variance, however, would be removed.


[/ QUOTE ]


If you just choose the hands where you get your money in to look at retrospectively, you get a very skewed image.

If you folded every hand that wasnt the stone cold nuts, that kind of 'analysis' would make you out to be a huge winner. But obviously that isnt the case.

It tells you something, but not much.

[SplawnDarts]

[ QUOTE ]

It's not complicated. Im not sure why you're having trouble understanding.


[/ QUOTE ]
I understand what you're saying just fine. It just happens to be wrong is all.
[ QUOTE ]

If you just choose the hands where you get your money in to look at retrospectively, you get a very skewed image.

If you folded every hand that wasnt the stone cold nuts, that kind of 'analysis' would make you out to be a huge winner. But obviously that isnt the case.

It tells you something, but not much.

[/ QUOTE ]

You're not following the method very well. Hands where resuts ARE expectation (such as folds) or where expectations cannot be accuratly calulated are left in the history unaltered. So the cost of folding while waiting for the nuts, for example, would be preserved.

[Abbaddabba]

Right.

So then i assume you would accept the standard deviation of a 5 hand sample too. is that correct?

because i have a 5 hand sample with a standard deviation of 0. will it suffice?

can i draw conclusions from that?

[ QUOTE ]
You're not following the method very well. Hands where resuts ARE expectation (such as folds) or where expectations cannot be accuratly calulated are left in the history unaltered. So the cost of folding while waiting for the nuts, for example, would be preserved.

[/ QUOTE ]

i said that it does tell you _something_.

just very little.

if you use it as a way of estimating winrates (which is what you're advocating) it fails miserably.

[SplawnDarts]

[ QUOTE ]
Right.

So then i assume you would accept the standard deviation of a 5 hand sample too. is that correct?

because i have a 5 hand sample with a standard deviation of 0. will it suffice?


[/ QUOTE ]
I call your bluff, obviously. You do not in fact have a random sample of 5 poker hands with a 0 standard deviation and a positive win rate. Assertions contrary to fact do not help your argument.
[ QUOTE ]

can i draw conclusions from that?


[/ QUOTE ]
No, because you made it up.

Also, had you passed statistics, you would know that such a smaple is insuficient in size to apply the central limit theorem, and as such confidence interval tests do not apply, so you cannot get fake results from your fake sample even if you tried.
[ QUOTE ]


[ QUOTE ]
You're not following the method very well. Hands where resuts ARE expectation (such as folds) or where expectations cannot be accuratly calulated are left in the history unaltered. So the cost of folding while waiting for the nuts, for example, would be preserved.

[/ QUOTE ]

i said that it does tell you _something_.

just very little.

if you use it as a way of estimating winrates (which is what you're advocating) it fails miserably.

[/ QUOTE ]

Actually, no, it doesn't. Even on large samples it gives you a MORE accurate estimate of the winrate than the sample itself. Try it some time and learn something [img]/images/graemlins/laugh.gif[/img]

[GiantBuddha]

This thread is awesome. And you should definitely play on the full moon, by the way.

[raju]

I thought this thread was going to be much more fun when i read this:


[ QUOTE ]
My strategy is simple (NO ITS COMPLICATED), but I am a selective maniac and use this style with my experience and the fact I can outplay almost everyone at the felt including good pros.

[/ QUOTE ]

NH Sir

[El_Hombre_Grande]

I usually tilt my tinfoil helmet ever so slightly to the left when this happens. I think, though, that it depends on whether you are North or South of the 45th parallel. If that doesn't work, get an MRI and see if any heavy metals are interfering with your ability to distinguish the flop from a distant memory of a re-run of high stakes poker. In most cases, 1200 hands is way too short term to determine if you are a losing player. In your case, I think I can draw some "down and dirty" conclusions, though.