How one guy turned his C&C skills into millions with online poker • Hacker News Board

Re: How one guy turned his C&C skills into millions with online poker

#41

Post by Eliezer » Mon, Mar 28, 2011, 7:59 AM UTC

danenania wrote: post #5

If he a.) plays at stakes where he can lose over a million in a couple days and b.) bases his decision to play someone (especially another pro) on emotion and recent short term variance, it's pretty much guaranteed he will end up broke eventually unless he actually has many many millions (like possibly in the hundreds if we're talking about enough to cover the swings of heads up PLO games at those stakes with Gus Han…

Lucky and unlucky streaks can and do last months to years for people who play thousands of hands per day.

This cannot possibly, possibly be true. The expected deviation goes as the square root of the sample size, which means that the percentage deviation goes as the inverse square root of the sample size, which means that if you play thousands of hands a day for a year, your deviation ought to be less than 1%.

Anything that persists over 100,000 samples is not luck.

Re: How one guy turned his C&C skills into millions with online poker

#42

Post by erikb » Mon, Mar 28, 2011, 8:40 AM UTC

mellowgold wrote: post #22

anagnorisis wrote: post #3

ps...check out the mathematic models/programs that play out and distill "variance", the "long term", and 'luck' in poker..It's remarkable stuff.

My first Django project was to create a simple variance simulator for poker ( http://www.evplusplus.com/poker_tools/variance_simulator/ ). The amount of luck involved in poker is truly tough to comprehend. Becoming a successful poker player not only requires you to be quite skilled, but you also need to run well at the beginning of your career.

Actually not. You didn't understand how luck works. Beeing successful in Poker is more about how you handle the bad streaks when they hit. Because they can hit often and hard. Everybody is running well from time to time.

Re: How one guy turned his C&C skills into millions with online poker

#43

Post by erikb » Mon, Mar 28, 2011, 8:46 AM UTC

Eliezer wrote: post #41

danenania wrote: post #5

If he a.) plays at stakes where he can lose over a million in a couple days and b.) bases his decision to play someone (especially another pro) on emotion and recent short term variance, it's pretty much guaranteed he will end up broke eventually unless he actually has many many millions (like possibly in the hundreds if we're talking about enough to cover the swings of heads up PLO games at those stakes with Gus Han…

Lucky and unlucky streaks can and do last months to years for people who play thousands of hands per day. This cannot possibly, possibly be true. The expected deviation goes as the square root of the sample size, which means that the percentage deviation goes as the inverse square root of the sample size, which means that if you play thousands of hands a day for a year, your deviation ought to be less than 1%. Anythi…

You mean it is quite unlikely that somebody has luck for as long as 100k hands or more. But the funny thing about "unlikely" is, that it occurs if you let a big group of people do the same stuff (like playing poker online). It is actually very, very likely that there are some (few) people who are hitting the luck box for much longer then just 100k hands.

1% seems to be a small number in your mind, I guess. Take one million people and if the chance to hit the jackpot is 0.1% you still get around one thousand people, who are lucky enough.

Maybe everthing changed, but in the early days of online poker, 6 or 7 years ago, we said that you need around 600k hands to have some(!) trust in your win rates (your Big Blinds per hundred hands, to be precise).

Re: How one guy turned his C&C skills into millions with online poker

#44

Post by sjm » Mon, Mar 28, 2011, 9:20 AM UTC

ianl wrote: post #29

Its very common for high level rts players to also play poker. During the time of starcraft broodwar before starcraft 2 launched there was not much north american/european interest in the gaming scene. Professional players use to play poker by day to make money and starcraft by night to practice for their next tournament.

In fact, one of the first foreign (non-Korean) StarCraft players to move to Korea went on to become a pro poker player (Bertrand 'ElkY' Grospellier: http://en.wikipedia.org/wiki/Bertrand_Grospellier)

From what I understand, many of the Korean pros get into online poker as well.

Re: How one guy turned his C&C skills into millions with online poker

#45

Post by ianl » Mon, Mar 28, 2011, 11:20 AM UTC

sjm wrote: post #44

ianl wrote: post #29

Its very common for high level rts players to also play poker. During the time of starcraft broodwar before starcraft 2 launched there was not much north american/european interest in the gaming scene. Professional players use to play poker by day to make money and starcraft by night to practice for their next tournament.

In fact, one of the first foreign (non-Korean) StarCraft players to move to Korea went on to become a pro poker player (Bertrand 'ElkY' Grospellier: http://en.wikipedia.org/wiki/Bertrand_Grospellier ) From what I understand, many of the Korean pros get into online poker as well.

One of the most prestigious non-korean tournaments is sponsored by Pokerstrategy.

http://www.teamliquid.net/forum/viewmessage.php?topic_id=205...

Re: How one guy turned his C&C skills into millions with online poker

#46

Post by khafra » Mon, Mar 28, 2011, 1:09 PM UTC

erikb wrote: post #42

mellowgold wrote: post #22

Earlier quoted context omitted.

My first Django project was to create a simple variance simulator for poker ( http://www.evplusplus.com/poker_tools/variance_simulator/ ). The amount of luck involved in poker is truly tough to comprehend. Becoming a successful poker player not only requires you to be quite skilled, but you also need to run well at the beginning of your career.

Actually not. You didn't understand how luck works. Beeing successful in Poker is more about how you handle the bad streaks when they hit. Because they can hit often and hard. Everybody is running well from time to time.

> You didn't understand how luck works.

That's a pretty bold statement to make to somebody who built a simulator. Can you point to any specific coding or algorithmic errors?

I think you misunderstood his point, which is that it's easy to burn through your starting bankroll, even if you play well.

Re: How one guy turned his C&C skills into millions with online poker

#47

Post by mellowgold » Mon, Mar 28, 2011, 2:44 PM UTC

erikb wrote: post #42

mellowgold wrote: post #22

Earlier quoted context omitted.

My first Django project was to create a simple variance simulator for poker ( http://www.evplusplus.com/poker_tools/variance_simulator/ ). The amount of luck involved in poker is truly tough to comprehend. Becoming a successful poker player not only requires you to be quite skilled, but you also need to run well at the beginning of your career.

Actually not. You didn't understand how luck works. Beeing successful in Poker is more about how you handle the bad streaks when they hit. Because they can hit often and hard. Everybody is running well from time to time.

"You didn't understand how luck works"

I have written quite a bit of poker software and played close to 700k hands of poker over the last 5 years so I understand how luck works better than most.

My point is, that if you imagine 10 hypothetical identical players (they make the same move in the same situations) with $1000 starting bankrolls, all with moderate a positive expectation of 1bb/100 hands many of those players will lose a lot at the beginning, get discouraged and give up playing poker in the first 10-20k hands they play because they ran poorly.

The people who were "lucky" and ran well at the beginning of there careers (like me) have a chance to grow a larger bankroll and will be less likely to go broke in the future as their risk-of-ruin decreases.

This is exactly what the simulator does. It takes identical players, under identical game conditions and looks at possible outcomes. You can see for yourself how wildly different the results can be for a player with a 1bb/100 win rate. (Note: the simulator of course is inherently flawed because games conditions and the mental condition of the player are not accounted for)

Re: How one guy turned his C&C skills into millions with online poker

#48

Post by danenania » Mon, Mar 28, 2011, 7:33 PM UTC

Eliezer wrote: post #41

danenania wrote: post #5

If he a.) plays at stakes where he can lose over a million in a couple days and b.) bases his decision to play someone (especially another pro) on emotion and recent short term variance, it's pretty much guaranteed he will end up broke eventually unless he actually has many many millions (like possibly in the hundreds if we're talking about enough to cover the swings of heads up PLO games at those stakes with Gus Han…

Lucky and unlucky streaks can and do last months to years for people who play thousands of hands per day. This cannot possibly, possibly be true. The expected deviation goes as the square root of the sample size, which means that the percentage deviation goes as the inverse square root of the sample size, which means that if you play thousands of hands a day for a year, your deviation ought to be less than 1%. Anythi…

The statistical math involved in determining standard deviation in poker is a lot more complex than you're giving it credit for. A hand itself is a very large set of potential outcomes, not just a win/lose scenario.

100,000 hands is nothing. Ask any long time pro who understands the game. I've personally known several who play a very consistent, disciplined style against a consistent level of opponents whose results would vary greatly across 100k hand sets, from being a huge winner to barely breaking even to losing even though they all were at least steady winners overall.

When you get in the 500k+ realm, it starts to get more reliable, but even streaks of this size are possible at the statistical fringes.

Re: How one guy turned his C&C skills into millions with online poker

#49

Post by Eliezer » Mon, Mar 28, 2011, 8:23 PM UTC

danenania wrote: post #48

Eliezer wrote: post #41

Earlier quoted context omitted.

Lucky and unlucky streaks can and do last months to years for people who play thousands of hands per day. This cannot possibly, possibly be true. The expected deviation goes as the square root of the sample size, which means that the percentage deviation goes as the inverse square root of the sample size, which means that if you play thousands of hands a day for a year, your deviation ought to be less than 1%. Anythi…

The statistical math involved in determining standard deviation in poker is a lot more complex than you're giving it credit for. A hand itself is a very large set of potential outcomes, not just a win/lose scenario. 100,000 hands is nothing. Ask any long time pro who understands the game. I've personally known several who play a very consistent, disciplined style against a consistent level of opponents whose results…

Still skeptical. What happens when you have pokerbots play against each other? Do they have similar variance across 100,000 hands?

Re: How one guy turned his C&C skills into millions with online poker

#50

Post by danenania » Mon, Mar 28, 2011, 8:56 PM UTC

Eliezer wrote: post #49

danenania wrote: post #48

Earlier quoted context omitted.

The statistical math involved in determining standard deviation in poker is a lot more complex than you're giving it credit for. A hand itself is a very large set of potential outcomes, not just a win/lose scenario. 100,000 hands is nothing. Ask any long time pro who understands the game. I've personally known several who play a very consistent, disciplined style against a consistent level of opponents whose results…

Still skeptical. What happens when you have pokerbots play against each other? Do they have similar variance across 100,000 hands?

Yes. Some of the first statistical analyses of poker I came across when I was learning were based on simulations using bots run by none other than Nate Silver (he used to be a poker pro). His conclusions were in line with what I've been saying, and these simulations were using full and 6-handed tables if I remember correctly. Heads-up has even more variance.