|
|
#31
06-26-2007, 11:53 AM
|
|||
|
|||
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
[ QUOTE ] So honestly, 2,000 hands tells you absolutely nothing. [/ QUOTE ] Honestly, if you do maths, you should run some numbers like raistlin's. You say "2000 hands, 50ptBB/100 winrate!" but we hear "I played poker online". [/ QUOTE ] I've played a lot more than 2,000 hands of poker online. 2,000 hands is since I got smart, focused on full ring NL and learnt what I was doing. 
|
|
#32
06-26-2007, 11:53 AM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
[ QUOTE ] [ QUOTE ] [ QUOTE ] What is your STD DEV for your sample size? [/ QUOTE ] Right, I'm going to run some checks on my hand history tonight and see. Honestly, not everyone's sitting there with PT, HUD and number coming out of their ears when they play poker. [/ QUOTE ] I understand that. But you claimed to be the math guy and yet you are making claims (granted "conservative") on the validity of win rates given a small sample size without having done the math to determine how large a sample size you actually needed. This is not how math is done. Since you are new to poker why don;t you start by asking some established posters what kind of a STD DEV they observe and then calculate how many hands you need to be within your chosen confidence interval on your win rate? As a base line, assuming a STDDEV of 40ptBB/100 hands you need something like 614,000 hands to be 95% confident your true winrate is with in 1ptBB/100 of the observed winrate. So honestly, 2,000 hands tells you absolutely nothing. [/ QUOTE ] I said I was a mathematician to give some context to the situation, not because I wanted to be mathematical in this thread. I could have asked what a typical S.D. was and gone away and done some calculations, but instead I asked for opinions. The I was amazed how conservative they were! Anyway, do you think 40ptBB/100 IS a common standard deviation? Over 2000 hands that would make the standard error in the mean WR 8.9ptBB/100. But say over those 2k hands I have been winning at $5/hr, which I make around 40ptBB/100 at 10NL (60 hand per hour). Then all of a sudden our confidence interval is about 20-60ptBB/100 (or $2.50 - $7.50). So you see why I say I am beating this game for $3/hour? btw, what makes you think I am new to poker? The only thing I am new to is thinking about poker as income. [/ QUOTE ] But the point is 2,000 hands is not enough to even start doing these calculations. If I said I played 10 hands and was beating the game for 40ptBB/100 wouldn't you laugh? Honestly, 2,000 hands isn't any more significant than 10. PS: you are completely ignoring the effects of some 6-10 BI downswings on your win rate, which is where you are going wrong in your thinking.
|
|
#33
06-26-2007, 12:01 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
I see why you are saying it. I don't believe its true.
Beating 10NL for 50ptBB/100? I see two possibilities: (1) You got dealt AA and KK a couple more times than you should have (which is unverifiable by the way, since you don't have pokertracker) and got paid off by donks. (2) You are among the best poker players ever. Over 2k hands I'm going to guess being a statistical outlier is a little more likely than #2. But, really there's two ways you can take this: Placate us by putting in another 28k hands at 10NL and establishing your winrate, or placating yourself by moving up and making money. Best of luck.
|
|
#34
06-26-2007, 12:02 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
[ QUOTE ] [ QUOTE ] [ QUOTE ] [ QUOTE ] What is your STD DEV for your sample size? [/ QUOTE ] Right, I'm going to run some checks on my hand history tonight and see. Honestly, not everyone's sitting there with PT, HUD and number coming out of their ears when they play poker. [/ QUOTE ] I understand that. But you claimed to be the math guy and yet you are making claims (granted "conservative") on the validity of win rates given a small sample size without having done the math to determine how large a sample size you actually needed. This is not how math is done. Since you are new to poker why don;t you start by asking some established posters what kind of a STD DEV they observe and then calculate how many hands you need to be within your chosen confidence interval on your win rate? As a base line, assuming a STDDEV of 40ptBB/100 hands you need something like 614,000 hands to be 95% confident your true winrate is with in 1ptBB/100 of the observed winrate. So honestly, 2,000 hands tells you absolutely nothing. [/ QUOTE ] I said I was a mathematician to give some context to the situation, not because I wanted to be mathematical in this thread. I could have asked what a typical S.D. was and gone away and done some calculations, but instead I asked for opinions. The I was amazed how conservative they were! Anyway, do you think 40ptBB/100 IS a common standard deviation? Over 2000 hands that would make the standard error in the mean WR 8.9ptBB/100. But say over those 2k hands I have been winning at $5/hr, which I make around 40ptBB/100 at 10NL (60 hand per hour). Then all of a sudden our confidence interval is about 20-60ptBB/100 (or $2.50 - $7.50). So you see why I say I am beating this game for $3/hour? btw, what makes you think I am new to poker? The only thing I am new to is thinking about poker as income. [/ QUOTE ] But the point is 2,000 hands is not enough to even start doing these calculations. If I said I played 10 hands and was beating the game for 40ptBB/100 wouldn't you laugh? Honestly, 2,000 hands isn't any more significant than 10. PS: you are completely ignoring the effects of some 6-10 BI downswings on your win rate, which is where you are going wrong in your thinking. [/ QUOTE ] 2000 hands no more significant than ten! What a ridiculous comment! I think the maths, that we are all so in love with, just proved how significant 2000 hands is. Not significant enough to give a winrate to the nearest BB/100, but significant enough to establish that someone who's crushing the tables is a big winner. Sure, a 6-10 BI downswing can occur, but so can a 6-10 BI upswing. That's why we're doing the maths isn't it? You can't insist on analysing the numbers, and then complain that the numbers don't take a downswing into account. Of course they do!
|
|
#35
06-26-2007, 12:05 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
I see why you are saying it. I don't believe its true. Beating 10NL for 50ptBB/100? I see two possibilities: (1) You got dealt AA and KK a couple more times than you should have (which is unverifiable by the way, since you don't have pokertracker) and got paid off by donks. (2) You are among the best poker players ever. Over 2k hands I'm going to guess being a statistical outlier is a little more likely than #2. But, really there's two ways you can take this: Placate us by putting in another 28k hands at 10NL and establishing your winrate, or placating yourself by moving up and making money. Best of luck. [/ QUOTE ] Well, I plan on doing the latter of course. Honestly people, these 10NL tables are bad. If you sit there and do some basic hand reading you kill them.
|
|
#36
06-26-2007, 12:08 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
Sure, a 6-10 BI downswing can occur, but so can a 6-10 BI upswing. That's why we're doing the maths isn't it? You can't insist on analysing the numbers, and then complain that the numbers don't take a downswing into account. Of course they do! [/ QUOTE ] Any 6-10 buyin downswings in you sample? No? Then no, it doesn't. Its like measuring earthquakes for a month, coming up with an average, and then saying that that average takes into account a 100-year event, even though there were no "big ones" in your sample.
|
|
#37
06-26-2007, 12:08 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
[ QUOTE ] [ QUOTE ] [ QUOTE ] [ QUOTE ] [ QUOTE ] What is your STD DEV for your sample size? [/ QUOTE ] Right, I'm going to run some checks on my hand history tonight and see. Honestly, not everyone's sitting there with PT, HUD and number coming out of their ears when they play poker. [/ QUOTE ] I understand that. But you claimed to be the math guy and yet you are making claims (granted "conservative") on the validity of win rates given a small sample size without having done the math to determine how large a sample size you actually needed. This is not how math is done. Since you are new to poker why don;t you start by asking some established posters what kind of a STD DEV they observe and then calculate how many hands you need to be within your chosen confidence interval on your win rate? As a base line, assuming a STDDEV of 40ptBB/100 hands you need something like 614,000 hands to be 95% confident your true winrate is with in 1ptBB/100 of the observed winrate. So honestly, 2,000 hands tells you absolutely nothing. [/ QUOTE ] I said I was a mathematician to give some context to the situation, not because I wanted to be mathematical in this thread. I could have asked what a typical S.D. was and gone away and done some calculations, but instead I asked for opinions. The I was amazed how conservative they were! Anyway, do you think 40ptBB/100 IS a common standard deviation? Over 2000 hands that would make the standard error in the mean WR 8.9ptBB/100. But say over those 2k hands I have been winning at $5/hr, which I make around 40ptBB/100 at 10NL (60 hand per hour). Then all of a sudden our confidence interval is about 20-60ptBB/100 (or $2.50 - $7.50). So you see why I say I am beating this game for $3/hour? btw, what makes you think I am new to poker? The only thing I am new to is thinking about poker as income. [/ QUOTE ] But the point is 2,000 hands is not enough to even start doing these calculations. If I said I played 10 hands and was beating the game for 40ptBB/100 wouldn't you laugh? Honestly, 2,000 hands isn't any more significant than 10. PS: you are completely ignoring the effects of some 6-10 BI downswings on your win rate, which is where you are going wrong in your thinking. [/ QUOTE ] 2000 hands no more significant than ten! What a ridiculous comment! I think the maths, that we are all so in love with, just proved how significant 2000 hands is. Not significant enough to give a winrate to the nearest BB/100, but significant enough to establish that someone who's crushing the tables is a big winner. Sure, a 6-10 BI downswing can occur, but so can a 6-10 BI upswing. That's why we're doing the maths isn't it? You can't insist on analysing the numbers, and then complain that the numbers don't take a downswing into account. Of course they do! [/ QUOTE ] Best of luck.
|
|
#38
06-26-2007, 12:11 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
Of course 2000 hands is more significant than 10 but in poker terms its still pretty insignificant. Certainly not enough to say if someone is a winner or loser let alone crushing the game.
After about 10k hands of NL10 I was beating it for 15PTBB/100 but then made a loss over my next 5000 hands. So do those 10k hands prove I'm crushing the game? Or do those 5k hands prove I'm a long term loser? The answer is neither!
|
|
#39
06-26-2007, 12:24 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
[ QUOTE ] Sure, a 6-10 BI downswing can occur, but so can a 6-10 BI upswing. That's why we're doing the maths isn't it? You can't insist on analysing the numbers, and then complain that the numbers don't take a downswing into account. Of course they do! [/ QUOTE ] Any 6-10 buyin downswings in you sample? No? Then no, it doesn't. Its like measuring earthquakes for a month, coming up with an average, and then saying that that average takes into account a 100-year event, even though there were no "big ones" in your sample. [/ QUOTE ] There's a two-and-a-half buy-in downswing
|
|
#40
06-26-2007, 12:28 PM
|
|||
|
|||
|
Re: Can I win $30/hour playing 25NL?
[ QUOTE ]
[ QUOTE ] [ QUOTE ] Sure, a 6-10 BI downswing can occur, but so can a 6-10 BI upswing. That's why we're doing the maths isn't it? You can't insist on analysing the numbers, and then complain that the numbers don't take a downswing into account. Of course they do! [/ QUOTE ] Any 6-10 buyin downswings in you sample? No? Then no, it doesn't. Its like measuring earthquakes for a month, coming up with an average, and then saying that that average takes into account a 100-year event, even though there were no "big ones" in your sample. [/ QUOTE ] There's a two-and-a-half buy-in downswing [/ QUOTE ] Besides, 6-10 BI downswings aren't "rare events" in the same way as massive earthquakes. The most you can lose on any one hand is your stack. A big 10BI downswing is made up of a number of smaller losses, which are all accounted for in the statistics.
|
 |
|
|
Linear Mode
