Computers Conquer Texas Hold'em Poker for First Time - Page 2 - Poker News

Two Plus Two Forums Poker News & Discussion News, Views, and Gossip

Computers Conquer Texas Hold'em Poker for First Time

Post Reply Subscribe

...

Page 2 of 16

1 2 3 4 5 6 7 12

Page 2 of 16

1 2 3 4 5 6 7 12

01-08-2015 , 07:37 PM

#26

TheDarkElf

veteran

Join Date: Nov 2010 Posts: 2,667

I heard if the bot takes a bad beat, it curses at the dealer.

Quote

01-08-2015 , 07:39 PM

#27

Hood

Carpal \'Tunnel

Join Date: Apr 2004 Posts: 6,887

Quote:

Originally Posted by 14756897412369

i assume game = hand?

correct

Quote

01-08-2015 , 07:52 PM

#28

Loctus

Carpal \'Tunnel

Join Date: Nov 2008 Posts: 11,156

Wait, does this graph mean that the computer will NEVER cap it preflop on the btn?? Any time it gets 3b pre it just calls?

Quote

01-08-2015 , 08:03 PM

#29

TChan

journeyman

Join Date: Mar 2005 Posts: 349

Quote:

Originally Posted by nburch

I'm kind of happy to see a resounding "Meh." We hoped everyone with a bit of knowledge would be aware that we hadn't somehow ruined the game of poker by finding an (almost) game theoretically optimal strategy, but I wasn't sure exactly how much credit to give people... It's oddly reassuring to see a bunch of people say "who cares?"

I think the more obvious question to me is, are you guys too good for money? Even starting from a modest bankroll, if you believe this much in your solution you should be now smashing the biggest limits. I'm always skeptical of people who claim solutions but are not willing to pony up against the best in the world. Shouldn't you be siphoning all the money out of the last few remaining HULHE specialists?

Quote

01-08-2015 , 08:07 PM

#30

CrazyLond

veteran

Join Date: May 2007 Posts: 2,930

Wow, am I correct that we can plug in any possible combination of action at a hulhe hand and it will return a perfect/gto solution? This is incredible.

Website down, looking forward to checking this thing out.

Quote

01-08-2015 , 08:08 PM

#31

nburch

newbie

Join Date: Jan 2015 Posts: 47

Quote:

Originally Posted by Loctus

Wait, does this graph mean that the computer will NEVER cap it preflop on the btn?? Any time it gets 3b pre it just calls?

Yes. At least if your opponent is perfect, it seems like it's a mistake to add that last raise. No guarantees if your opponent is not perfect, or you end up making mistakes later on in the game...

It caps .06% of the time with 9T suited - most likely part of the noise that makes it "essentially" solved and not just solved.

Quote

01-08-2015 , 08:10 PM

#32

Frankie Fuzz

grinder

Join Date: Aug 2013 Posts: 534

Quote:

Originally Posted by TChan

It seems like you are making one or more of the following assumptions, none of which are likely to be true:

1) It is easy for a human to simply observe the bot and closely replicate its play.

2) They are willing to break site rules and let their bot play on an online poker site, discrediting their academic department.

3) Opponents will continue playing them for a considerable amount of time after observing that they play well.

Quote

01-08-2015 , 08:11 PM

#33

NoQuarter

old hand

Join Date: Dec 2014 Posts: 1,874

So, how long before the darkside nerd hacks steal this program, incorporate it into their own bot, then rape online poker?

RIP online poker IMHO

Quote

01-08-2015 , 08:13 PM

#34

14756897412369

centurion

Join Date: Jan 2015 Posts: 102

Quote:

Originally Posted by Frankie Fuzz

It seems like you are assuming one of 2 things, neither of which are likely to be true:

1) It is easy for a human to simply observe the bot and closely replicate its play.

2) They are willing to break site rules and let their bot play on the site, which would discredit their academic department as well.

3) anyone would play against them

Quote:

Originally Posted by NoQuarter

So, how long before the darkside nerd hacks steal this program, incorporate it into their own bot, then rape online poker?

RIP online poker IMHO

yeah because there is just so much action for hulhe...

Quote

01-08-2015 , 08:13 PM

#35

Loctus

Carpal \'Tunnel

Join Date: Nov 2008 Posts: 11,156

Capping pre in HULHE confirmed non-GTO.

Quote

01-08-2015 , 08:15 PM

#36

nburch

newbie

Join Date: Jan 2015 Posts: 47

Quote:

Originally Posted by TChan

As someone already brought up - GTO doesn't necessarily mean you'll be smashing people. It's very (very) good at not making mistakes, but not as good as pros at making someone pay for their mistakes.

So... against any comer, it would be a good bet to say the bot wins, but it could be quite a few hands to actually move it from a better-than-even coin flip. I _DO_ think a lot of poker players underestimate just how noisy money lines are: there's awareness that's it noisy, but I'm pretty it's actually worse than (many) people think.

Quote

01-08-2015 , 08:16 PM

#37

chinagambler

veteran

Join Date: Jun 2011 Posts: 2,572

Quote:

Originally Posted by TChan

There's a difference between finding the solution and actually memorizing it.

Quote

01-08-2015 , 08:17 PM

#38

Joe Knott

grinder

Join Date: Nov 2008 Posts: 431

Quote:

Originally Posted by Loctus

Wait, does this graph mean that the computer will NEVER cap it preflop on the btn?? Any time it gets 3b pre it just calls?

Are you colorblind? He caps both KK and QQ and even some more, 0.02 % of the time

I'm surprised that there exists a strategy in the support that is played with almost zero frequency though :O

Last edited by Joe Knott; 01-08-2015 at 08:23 PM.

Quote

01-08-2015 , 08:43 PM

#39

JimAfternoon

Carpal \'Tunnel

Join Date: Mar 2011 Posts: 6,903

This is really cool.

Quote

01-08-2015 , 09:34 PM

#40

ArtyMcFly

Carpal \'Tunnel

Join Date: Dec 2014 Posts: 13,256

Congrats to the team, but this is really bad news for poker as a long-term profession.

Quote:

Originally Posted by pocket_zeros

Uses 200 computer nodes each consisting of 24 2.1Ghz AMD cores, 32GB of ram, and 1TB disk

Entire solution will run on a graphene smartwatch in a few years.

InB4 a simcard the size of a fingernail wins a bracelet. #RobotTakeover

Quote

01-08-2015 , 09:37 PM

#41

14756897412369

centurion

Join Date: Jan 2015 Posts: 102

good players will find a way to adapt. those that don't will be left behind like always.

Quote

01-08-2015 , 09:43 PM

#42

Loctus

Carpal \'Tunnel

Join Date: Nov 2008 Posts: 11,156

Quote:

Originally Posted by 14756897412369

good players will find a way to adapt. those that don't will be left behind like always.

The only adaptation that can come from this is really that online HULHE for money dies, like online chess for money did.

Not much change compared to today, in other words

Quote

01-08-2015 , 09:53 PM

#43

sthief09

Josh.

Join Date: Feb 2004 Posts: 20,654

Quote:

Originally Posted by TChan

I think the more obvious question to me is, are you guys too good for money?

they're academics. in other words, yes

Quote

01-08-2015 , 10:13 PM

#44

Garick

Oberbiergenießer

Join Date: Dec 2007 Posts: 26,514

GTO means "not exploitable" which means "doesn't care how you respond." That does not mean "makes a ton of money." GTO is in no way "optimal" as poker player describe it (makes the most money). It is only "optimal" in that no matter how good you are you can't make a GTO strategy lose a ton of money. It is very likely that GTO in a fair zero sum poker game (no rake) is exactly neutral EV against everyone, from Joe Donk to Phil Ivey.

It is also possible (likely, imo) that in a raked game, GTO is a guaranteed slow loser.

Quote

01-08-2015 , 10:18 PM

#45

14756897412369

centurion

Join Date: Jan 2015 Posts: 102

so even if i shove every hand in nlhe or bet/raise every street in lhe, it's still possible that gto is neutral ev vs me?

Quote

01-08-2015 , 10:20 PM

#46

CrazyLond

veteran

Join Date: May 2007 Posts: 2,930

In a poker game perfectly unexploitable play will beat everyone who is not playing an identical strategy. In some games you are correct that it will break even but poker is not one of them. In poker the further you are from unexploitable the more you will lose vs. this bot.

Just think about some absurd strategy like open folding every button. Obviously the bot would crush this strategy.

Now the bot won't win as much from a bad player as a human who can adjust to exploit might but it will beat everyone to some extent.

Quote

01-08-2015 , 10:33 PM

#47

News777

adept

Join Date: Sep 2007 Posts: 765

If you could take the computer which is illegal in the casino to play the Heads-Up poker video game you probably could destroy it.

Quote

01-08-2015 , 10:43 PM

#48

kaby

Carpal \'Tunnel

Join Date: Jan 2007 Posts: 6,230

Congratulations guys, impressive job.

Quote

01-08-2015 , 10:45 PM

#49

Wilbury Twist

Carpal \'Tunnel

Join Date: May 2008 Posts: 6,153

Quote:

Originally Posted by Garick

Thank you... I was about to ask if GTO is necessarily the most profitable in HULHE.

Quote

01-08-2015 , 11:08 PM

#50

RonMexico

Pooh-Bah

Join Date: Feb 2006 Posts: 4,323

Quote:

Originally Posted by Garick

It is very likely that GTO in a fair zero sum poker game (no rake) is exactly neutral EV against everyone, from Joe Donk to Phil Ivey.

It is also possible (likely, imo) that in a raked game, GTO is a guaranteed slow loser.

ffs no

Quote

Page 2 of 16

First

1 2 3 4 5 6 7 12

Last

Post Reply Subscribe

...

Page 2 of 16

First

1 2 3 4 5 6 7 12

Last