PDA

View Full Version : Short Run / Long Run

DiceyPlay
11-26-2004, 07:21 PM
I've heard it said and it seems reasonable that only those players who play consistently 30+ hours a week for an extended period of time (like more then 10 years) get to the long run.

Assuming that is true, isn't it also true that most players only experience short term statistical anomolies?

So how do you know if you're playing well?

How do you come to grips with the fact that your experience isn't necessarily representative of your expectation (it only most likely represents what you should expect)?

You just have to keep playing and hope you don't end up one of the statistical abberations of negative results.

Because if you don't get to the long run, your career results are only a statisical abberation.

I'm leaving to play in a half hour.

Any thoughts appreciated.

-DP

gaming_mouse
11-26-2004, 07:44 PM
who play consistently 30+ hours a week for an extended period of time (like more then 10 years) get to the long run.

DiceyPlay,

It looks like you have the opposite problem from most people: you are overestiming the number of hands required for a statistically significant result. Most poeople underestimate.

Ten years is WAY OFF, though. 30K hands is usually sufficient to see if you are winning or losing (though often, 10K is enough). In any case, if you are playing 8 hours a day online, you should know within a month.

Don't imagine "the long term" to be some mysterious, nebulous entity which can never be attained. You just estimate your win rate and a standard error. The standard error gets smaller the more hands you play. Eventually (unless you are a break even player) you will have evidence that your win rate is positive or negative.

You can search the site to find the appropriate formulas for doing the calculations.

gm

DiceyPlay
11-27-2004, 03:51 AM
May be I am off with my 10 year figure. I was hoping to get some insightful conversationn about this interesting topic. But you pretend to be an expert and trivialize my original thought.

"(though often, 10K is enough)" - where did this come from? I'm sure you're an expert in these matters, that's why they're trivial, right? &lt;do you detect sarcasm?&gt;

Playing on the internet is not the same as live play. On the internet people multi-table, don't pay attention (but you don't know if they're paying attention or not). Both of these factors skew results. I'm sure there are plenty of other factors that taint results on the internet.

You don't estimate your win rate. Your win rate is your win rate. You estimate your population standard deviation (something you never attain) with your sample standard deviation. But it appears the technical details are beneath you.

What I'm saying is that what you experience can only suggest whether your win rate is positive or negative. You can only be confident to some degree, you can NEVER be certain.

I'm not imagining anything. Statistics is far more significant then you give it credit. If you can't appreciate that and ponder how it fits in with your own experience playing, I doubt you can beat the game unless you're one of these guys with natural talent - I'm not.

The formulas don't help me think about the concepts and there are better places (more efficient) to find technical reference then this site.

Anyway ... if anyone has some insight or constructive criticism, I'd like to hear it. I just don't like it when people seem to only want to flex their noodle.

-DP

jtr
11-28-2004, 10:44 AM
Jeez, guy. Way to invite further responses. Thought of trying decaf?

I don't mean to inflame you, and I am completely sympathetic to your general point. Yes, those of us who don't play enough hands may well be living in statistical anomaly land. I won't enter the fray on exactly how many hands constitutes the long run, but it's certainly true that many of us don't play enough hands in a month or a year to have any sort of decent estimate of our long-run expected win rate. I remember when I first started working with some simple simulation code of my own for poker, and being alarmed to find that two players on the exact same strategy could play 100,000 hands or more and still have very different results.

But I find this fact to be up there with a lot of unpleasant things in life, like death and taxes. Not much we can do about it (except play more hands, of course). What kind of answer were you looking for?

I guess your key question is how to know whether you're playing well, despite the fact that your short-term results are a terribly flawed measure of the quality of your play. I think an appreciation of this problem is what drives the "post hands" mantra that one finds on these forums. You have to make sure that your strategic thinking is sound, by checking it out with others who think hard about the same problem. If there's a general consensus that you're making all the right moves in your posted hands, but you're losing money, then you should be confident to reject the short-term result and stick with your strategy.

I think that's all we can do.

topspin
11-30-2004, 04:11 AM
[ QUOTE ]
I just don't like it when people seem to only want to flex their noodle.

[/ QUOTE ]

[ QUOTE ]
Statistics is far more significant then you give it credit

[/ QUOTE ]

I hardly think gaming_mouse is just "flexing his noodle".

The formulas he is referring to are based on statistics. They give you a way of establishing confidence intervals for your win rate estimate, and hence tell you how many samples you need before you can start trusting your numbers. You can plug in your standard deviation and calculate for yourself how many hands you need, and figure out if for your play you're likely to reach the "long term".

What sort of answer were you expecting?

DiceyPlay
11-30-2004, 01:07 PM
Thanks jtr. Thanks for responding (after my ceffeine induced post) and thanks for your response.

The thought of "two players on the exact same strategy could play 100,000 hands or more and still have very different results" can be depressing/disheartening and almost defeating. One just has to hope they aren't in the "getting the shaft" side of results.

I like what you suggest. Post hands, general concensus, ignore results.

May be some other things you can do are:

Study to build confidence that the moves you make are correct regardless of the result. You can keep a journal and reflect upon your past play to help you with this.

Be selective of the games you sit in and make sure you always have an edge (i.e. a soft spot in the lineup to exploit or ???). Otherwise, don't play.

Don't play / Stop playing when tired.

Take breaks to make sure you don't lose your mental acuity.

Thanks,

-DP

ZeeBee
12-01-2004, 08:28 PM
[ QUOTE ]
I was hoping to get some insightful conversationn about this interesting topic. But you pretend to be an expert and trivialize my original thought.

[/ QUOTE ]
You know, I've read and re-read gaming_mouse's reply and I can't see anything other than helpful pointers in it. You are being waaaay over sensitive.

He's right too - search the site and you'll find how to calculate a confidence interval for your "true" winrate using your standard error. I suspect you'll have to search for it because I doubt gaming_mouse will take the time to look it up for you given your response.

Chill out and have fun.

ZB