People and their money. - Page 2

jtr · #11 10-01-2007, 10:40 AM

[ QUOTE ]

I have over 30 million hands in PokerTracker and when i parse out those people who have less than 10k hands, i still find more than 30% are total winners. I believe 5-10% is too low a guess.

[/ QUOTE ]

What was the question you were trying to answer? If it was the percentage of winning players amongst those who play a lot of online hands, then sure, the answer might be different from 10% or so. But that wasn't the question.

Furthermore, think for a second about the inherent variance in a 10K hand sample. Assume whatever winrate you want and a standard deviation of 35 PTBB/100; that means that the standard deviation of observed winrates after 10K hands will still be 3.5 PTBB. (10K hands means a hundred 100-hand blocks, so your sample size is 100, and the standard error is equal to SD over the square root of n.)

Now, let's suppose that *all* the people you have 10K hands on are actually -3.5 PTBB/100 losers. In this case we would still expect to see about 16% of them showing up as winners.

I think you're underestimating the certainty you get from 10K hand samples, or at least missing the fact that simply collecting a lot of 10K hand samples doesn't make each sample any less noisy.

SellingtheDrama · #12 10-01-2007, 01:45 PM

Personally I've yet to see the point in trying to determine a true winrate. Unless you are a bot incapable of learning/adapting, it is pretty safe to say that over any 100k (probably a lot less) hand stretch, you will learn something and adapt your game - which will change your winrate (hopefully for the better).

Also, since I generally play live, I don't get anywhere near the hands or hours in that some people do. I've learned to settle for 'probably a winner' and similar assessments, both of myself and others.

jtr · #13 10-01-2007, 02:39 PM

Sounds like a very reasonable approach to me.

KurtSF · #14 10-01-2007, 02:45 PM

[ QUOTE ]

I have over 30 million hands in PokerTracker and when i parse out those people who have less than 10k hands, i still find more than 30% are total winners. I believe 5-10% is too low a guess.

[/ QUOTE ]

Selection bias. Winners will tend to play more hands than losers and therefore show up in your >10k hand sampling more often than losers. Your data is essentially meaningless.

KurtSF · #15 10-01-2007, 03:31 PM

And regarding the OP, 2 years ago when I had some extra money I started with a $500 bankroll and worked hard at learning the game and grinding. As soon as I was a winning player I paid myself back the $500 and have only played on winnings since, moving up in levels when my bankroll allowed. I treat it like a business, and keep my bankroll totally separate from my income and living expenses. I have about $50 in discretionary income each month (what's left from my paycheck after rent and food), so maintaining a bankroll and practicing good bankroll management is crucial to keeping me in the games.

Currently my bankroll is split up with more than half in a high interest savings account, about $1k on deposit at my regular cardroom, $1k in cash at home for playing whatever else, and about $1k online if I want to play there.

KhalynYohrk · #16 10-01-2007, 03:57 PM

What % response do you think we'd get if we asked people if they were winners in the long run. Denial is a powerful thing. 30-40%?

SellingtheDrama · #17 10-01-2007, 04:07 PM

[ QUOTE ]
What % response do you think we'd get if we asked people if they were winners in the long run. Denial is a powerful thing. 30-40%?

[/ QUOTE ]

I'm pretty sure that 90% of poker players think they are unluckier than everyone else at their table.

KurtSF · #18 10-01-2007, 04:19 PM

[ QUOTE ]
What % response do you think we'd get if we asked people if they were winners in the long run. Denial is a powerful thing. 30-40%?

[/ QUOTE ]

Over 50% I would guess.

jtr · #19 10-01-2007, 05:00 PM

Just like driving, everyone is better than average.

Ganjasaurus Rex · #20 10-01-2007, 07:21 PM

[ QUOTE ]

I think you're underestimating the certainty you get from 10K hand samples, or at least missing the fact that simply collecting a lot of 10K hand samples doesn't make each sample any less noisy.

[/ QUOTE ]
I'm aware of the noise. I'm also aware that there is no fixed threshold above which variance disappears. Furthermore, i'm aware that the best indicator of the population mean is the sample mean, however small the sample might be. If you believe the sample mean to be different, then the onus is on you to provide evidence to the contrary.