How long is the long-run in poker? How many hands do you need to be certain whether you are a winning or loser player?

I understand that the game evolves, but let's just say the skill level stays the same for all players for a period of time, how many hands do you need to make a good assumption?

Also, how long can a downswing last? At what point does a downswing just really mean you are a bad player?

Sorry for all the questions, just curious as to what people define as short-run and long-run in poker and the classifications of swings.

In NLHE, generally around 50K hands at the same stakes and against the same type of opponents, is enough to establish a pretty reliable winrate and standard deviation for your play. But during that time your play and opponent's play could be evolving too (as you mentioned), so you will never know exactly.

You can determine how good a player you are much faster if you do more than look at your raw profit or loss. For one obvious example, if you get all-in with the odds in your favor that expected gain should be credited to your account for evaluation purposes, whether or not you win. Similarly if you get all-in with negative EV, that counts against you, even if you get lucky.

More generally, you should do things for reasons. If you bet to make someone else fold, and they do, that testifies to your skill. If you limp with a small pair because you expect a multiway pot, and four other players see the flop, that's a success even if you don't get trips and fold immediately. If someone beats your two pair with a straight that you didn't notice was possible, that's bad, worse than if you had lost to a hand you had considered and decided to bet against.

Subjecting your game to rigorous evaluation is both the way to improve and the way to learn how good you are. Of course, if your actual results are consistently below what you think they should be based on your analysis, you have to accept that your analysis is flawed.

It depends on what game you are playing but in NLHE which i would imagine you are referring to, then 50k like spacebidder mentioned is probably really reasonable

50k hands seems a lot for live games though.

It is a lot and might take 2 years, but there is no reason to need a smaller sample size in a live game if you want the same confidence in your result.

Pretty sure the long run is infinite, however there is minor milestones on the way that give you some sort of idea. Anyone can go on a 5 million hand downswing/upswing.

If you're on a 5 million hand downswing, it might be time to man up and admit that you're actually just not a very good poker player...

this is nonsense










barry

http://www.evplusplus.com/poker_tool...nce_simulator/

Should help you, even on 1M hands and a 4bb/100 winrate, expected winnings can be very different

Statistics aren't really my bag, but I do know there is something called a 'statistically significant sample' for a particular population.

There are 8 x 10^68 ways to shuffle a deck of cards. Multiply that by the number of seats at the table, and I think that represents the total "long run".

Can someone who is more learned on the subject tell me what a statistically significant sample would be based on that population?

I believe the answer to that is the answer to "how long is the long run?"

But even if it's something small like one-one hundred millionth of one percent; that's still more hands than you can play in ten lifetimes.

So I doubt the "long run" even exists.

You're mixing up two things here a bit. The number of combinations of cards and seats isn't particularly relevant, because

a) the effective number of combinations is way lower because from your point of view, many combinations are functionally equivalent. For example in a 10 player game, there are only 25 cards in play, not 52. Then there are situations that are functionally the same such as Kh5s vs Kc5d etc.

b) the actual number of combinations doesn't really matter as much as the distribution of outcomes. For example. Say we're going to play a game where I write down a number and without looking at it, you guess whether it's odd or even. The sample space is infinite - but the "long run" is very short because you have a 50% chance of guessing. It's quite easy to calculate how often you'll guess 5 times wrong in a row, or how often you'll get 10 wrong in a row out of 1000 guesses, etc.

The worst case for variance is actually 50/50. There is a formula for estimating how many iterations you need before you have, say, a 95% chance of being within 1% of the "true" outcome. I have it here:
http://rustybrooks.com/poker/monte_c...imulation.html
it's a lot smaller than you think.

The long run is forever.

long run is infinity. what you get with bigger samples is higher accuracy and smaller probability that your negative winnings are just a downswing or that your positive winnings are just an uspwing. it takes an awful lot of time to make this probability small enough to have some faith in it, but usually whether you're a winning or a losing player can be deduced from your playstyle faster than that

The long run is not infinity. The long run is not 50k hands. The long run isn't a phrase that actually means anything other than what you want it to mean.

This is why statisticians use phrases that have specific means. An example is what you'd find in my link above. For example: "How many hands do you need to play to have a 95% confidence that your results could not be obtained by a player who is breakeven or worse"

Instead of 95 you can use 99 or 99.5 or whatever (but never 100% - there is no such thing)

So you need to figure out what you want to know. Instead of saying "how many hands until I'm certain I'm a winning player" ask "how many hands until I can be 95% certain I'm a winning player". Even this is really a bit of an approximation, since you as a player and your competition are always in flux to some extent, but, it's a great deal better than nothing.

I used to hurry a lot. I used to worry a lot. I used to stay out till the break of day. Oh, that didn't get it, it was high time I quit it, I just couldn't carry on that way. Oh, I did some damage, I know it's true. Didn't know I was so lonely , till I found you. You can go the distance.

We'll find out in the long run.

This is an unnecessarily complex and esoteric answer to a very informal question posed by the OP. The OP really asked two questions:

A: What is "the long run?"
B: How many hands does it take for one to be certain that he/she is a winning poker player?

The answer to A is undoubtedly infinity. The answer to B is what you tried to explain, which is that you can't really answer that question without first establishing a set of parameters (Ie; what definies a 'winning poker player') within which to make that judgement.

"Infinity" might be a short and sweet answer, but it doesn't help anyone, really. And I'm not sure I really agree that it's the right answer. There is not likely to be any formal answer to "what does the long run mean"

I agree, if the long run is as long as infinity then it is the same as if it doesn't exist since no one gets to live forever.

I must say here, since i also struggled with the term, that when you come across this concept of the long term in almost every training site, book etc. etc. a point comes where you actually do need a solid definition of it if one can't be given then it should not be used the way it is.

The term "long run is" used too much to form, or validate, a very specific kind of play, while at the same time it is an extremely vague concept, it doesn't feel right.

"Do as i say and in the long run you will win" OK fine, but give me a number, be more specific please so i can know when to sue you if i don't.

You can do all the math in the world but when you run bad there is absolutely nothing you can do about it. I was one outed once and two outef 13 times yesterday as an 83% favourite preflop and i got it in pre or on the flop flop as still huge favourite only to have them hit miracle after miracle all in a short 6 hour period. So i for one think you need a fairly significant sample to allow for variance and other typical online nonsense.

doesnt matter if you look left and right when crossing the street if there is a meteor aimed at your head, but that doesnt mean you shouldnt look

I mean I assume long-run is "infinity" but yes the bigger the sample the more dialed-in you are into how your game really is.

This may be a cash game vs. tournament situation but should weighting even come into poker statistics ie Affleck hand costing him millions of dollars or taking a shot at 5/10 when rolled for 1/2? I guess this is why BRM is very important.

I'm going to chime in here with one more thing.

You need to consider the QUALITY of the sample you're evaluating.

In other words, if you decide 100K hands is the magic number, did you play your A game for all 100k hands? And was your A game for the first 10K hands as good as it was on the last 10K hands?

Regardless of what teh consensus here believes the "long run" is, if you're not a significantly better player by the end of it, then you've done something horribly wrong. So whatever statistics you come up with for a sample size, win rate, etc, is going to be wrong also.

Bottom line: every second that you spend worrying about win-rates, sample sizes, or stats, is a second that you didn't spend getting better at poker. Nothing you can calculate from past experiences can help you accurately predict future events. Poker just isn't like that.

Just play the stakes you're bankrolled for, and try to get better. Everything else is an EPIC waste of time.

Probably. There are statistical models that take a crack at stuff like this but they are complicated and problematic.

But the real point of the above is, if you get better over 100k hands, then your statistical analysis will *underestimate* your win rate, so you can still be relatively assured it is that or better.

The long run is not a time period.

The law of large numbers simply says that as you perform more trials the average result will get closer to the expected value. This is true for any time period whether it be 10 hands (closer than 1) or 10 million (closer still). The "long run" when speaking of probability, is just a casual reference to this property.

A more useful definition can be found in the world of advantage blackjack. They use a statistical term "N₀" (N zero) which is the time in hands necessary for the expected winnings to equal one standard deviation. The term is specifically used as a working definition of "the long run" and was created for that reason.

N₀ can be thought of as the time for the non-random component to begin dominating the random component (to be nitty this is true 84% of the time when using 1 SD).