Two Plus Two Publishing LLC
Two Plus Two Publishing LLC
 

Go Back   Two Plus Two Poker Forums > >

Notices

Mid-High Stakes Limit Discussions of mid-high stakes limit Texas Hold'em

Reply
 
Thread Tools Display Modes
Old 08-24-2020, 05:50 PM   #301
maka2184
adept
 
Join Date: Oct 2006
Location: Midwest
Posts: 999
Re: 2020 HI content

Quote:
Originally Posted by hardinthepaint View Post
Being a recreational player doesn't make your views any less valid. But you have provided 0 justification for your position. I am fine with "agreeing to disagree"; I don't care about convincing you.



Hard pass on that.
Good luck have fun

PS: I'm a TAG fish always available for donation if we ever run into each other live LHE Midwest or Commerce post vaccine.

OT: Sean Snyder twitter 2020
maka2184 is offline   Reply With Quote
Old 08-26-2020, 12:37 AM   #302
Unguarded
Pooh-Bah
 
Join Date: Apr 2006
Location: worrying what PokerBob thinks of me
Posts: 3,558
Re: 2020 HI content

Maka, the story of Cepheus is really quite fascinating just based on what little I know. As I understand the algorithm, they basically just taught it the rules of HUHU LHE and had it play against itself a zillion times until it was ridic close to the solution for the game. Using this "simple" technique, they got far closer to the solution than they had in almost a decade of working on the project. For example, Polaris was the first HUHU LHE bot to defeat a top human (HossTBF). But they estimated that Polaris was something like .85 BB/100 from a perfect GTO solve. I suspect that the 2008 version of Polaris was still strong enough to beat today's top players. But interestingly, I suspect that today's top humans would significantly outperform Polaris against a recreational player since their exploitative arsenal is so powerful while Polaris just spams a static strategy.

What I find most interesting is that the algorithm used to build Cepheus sounds almost exactly like the one used to build AlphaGo and AlphaZero, the bots that made headlines by dominating Go and chess respectively. Those 2 bots are considered revolutionary, so it is super cool that the Cepheus team managed to get there 5 years before AlphaZero and AlphaGo. If I were less lazy, I would check to see if some members of the Cepheus team worked on the chess and Go bots.
Unguarded is offline   Reply With Quote
Old 08-26-2020, 06:08 AM   #303
maka2184
adept
 
Join Date: Oct 2006
Location: Midwest
Posts: 999
Re: 2020 HI content

Quote:
Originally Posted by Unguarded View Post
Maka, the story of Cepheus is really quite fascinating just based on what little I know. As I understand the algorithm, they basically just taught it the rules of HUHU LHE and had it play against itself a zillion times until it was ridic close to the solution for the game. Using this "simple" technique, they got far closer to the solution than they had in almost a decade of working on the project. For example, Polaris was the first HUHU LHE bot to defeat a top human (HossTBF). But they estimated that Polaris was something like .85 BB/100 from a perfect GTO solve. I suspect that the 2008 version of Polaris was still strong enough to beat today's top players. But interestingly, I suspect that today's top humans would significantly outperform Polaris against a recreational player since their exploitative arsenal is so powerful while Polaris just spams a static strategy.

What I find most interesting is that the algorithm used to build Cepheus sounds almost exactly like the one used to build AlphaGo and AlphaZero, the bots that made headlines by dominating Go and chess respectively. Those 2 bots are considered revolutionary, so it is super cool that the Cepheus team managed to get there 5 years before AlphaZero and AlphaGo. If I were less lazy, I would check to see if some members of the Cepheus team worked on the chess and Go bots.
Thank you very much Unguarded!

Think reopening my MS word to add this to 2+2 epic post(s)

OT: Won once at above 1k post qurantine playing under 6 hours

Hopefully, my 5k downswing this year will break even during Fall 2020
maka2184 is offline   Reply With Quote
Old 08-26-2020, 11:57 AM   #304
Munga30
veteran
 
Munga30's Avatar
 
Join Date: Sep 2002
Location: Sacking up and saying thank you.
Posts: 2,408
Re: 2020 HI content

Quote:
Originally Posted by Unguarded View Post
Maka, the story of Cepheus is really quite fascinating just based on what little I know. As I understand the algorithm, they basically just taught it the rules of HUHU LHE and had it play against itself a zillion times until it was ridic close to the solution for the game. Using this "simple" technique, they got far closer to the solution than they had in almost a decade of working on the project. For example, Polaris was the first HUHU LHE bot to defeat a top human (HossTBF). But they estimated that Polaris was something like .85 BB/100 from a perfect GTO solve. I suspect that the 2008 version of Polaris was still strong enough to beat today's top players. But interestingly, I suspect that today's top humans would significantly outperform Polaris against a recreational player since their exploitative arsenal is so powerful while Polaris just spams a static strategy.

What I find most interesting is that the algorithm used to build Cepheus sounds almost exactly like the one used to build AlphaGo and AlphaZero, the bots that made headlines by dominating Go and chess respectively. Those 2 bots are considered revolutionary, so it is super cool that the Cepheus team managed to get there 5 years before AlphaZero and AlphaGo. If I were less lazy, I would check to see if some members of the Cepheus team worked on the chess and Go bots.
I have no real insight into the Cepheus or AG/AZ algorithms, but this video discusses a fascinating approach for developing a strong player.

Last edited by Munga30; 08-26-2020 at 12:04 PM.
Munga30 is offline   Reply With Quote
Old 08-26-2020, 01:14 PM   #305
maka2184
adept
 
Join Date: Oct 2006
Location: Midwest
Posts: 999
Re: 2020 HI content

Quote:
Originally Posted by Munga30 View Post
I have no real insight into the Cepheus or AG/AZ algorithms, but this video discusses a fascinating approach for developing a strong player.
Thank you.

Will watch after maket close.


*Very OT ON NBA HC 2020+*
Which team would be most +EV to hire Dantoni?

Thinking Chicago Bulls to help recruit?


Adrian Wojnarowski (@wojespn) tweeted at 10:33 AM on Wed, Aug 26, 2020:
Rockets coach Mike D’Antoni — in the final year of his contract — will be an Indiana target should he become available, sources tell ESPN.
(https://twitter.com/wojespn/status/1...029737984?s=03)

Get the official Twitter app at https://twitter.com/download?s=13
maka2184 is offline   Reply With Quote
Old 09-03-2020, 05:46 PM   #306
___1___
old hand
 
Join Date: Dec 2004
Location: __________
Posts: 1,552
Re: 2020 HI content

Is there any proof that would reasonably convince that the majority of the players on Bodog 10/20 and 30/60 are bots (house bots or otherwise)? Does that proof not exist almost by definition on an anonymous site?
___1___ is offline   Reply With Quote
Old 09-06-2020, 12:28 AM   #307
AlanBostick
Carpal \'Tunnel
 
AlanBostick's Avatar
 
Join Date: Sep 2002
Location: Old Man Red Bull (he/him)
Posts: 10,477
Re: 2020 HI content

Quote:
Originally Posted by ___1___ View Post
Is there any proof that would reasonably convince that the majority of the players on Bodog 10/20 and 30/60 are bots (house bots or otherwise)? Does that proof not exist almost by definition on an anonymous site?
Doug Polk offers four indicators of possible bot play:


1. The player takes about the same amount of time to act every hand.
2. The player doesn’t answer to moderator in the chat or to an alert.
3. The player plays for unreasonable amounts of time.
4. The player plays an unreasonable amount of tables.

Unfortunately, only the first indicator is something that a player on Ignition could be in a position to observe. There is no chat on Ignition (there may well be on Bodog, but I can't observe it from Ignition). There is no way to tell how long a player has been playing, unless you, too, have been playing with them for a long time at the same table. The software limits cash tables to four, so a single account can play in only four cash games at a time -- but I understand that tournament tables are not limited in this way. In any event, players have no way to determine how many tables another player is playing.

Ignition does have the feature that (twenty-four hours after the player leaves the cash game or after a tournament ends) they can download all the hand histories, complete with revealed hole cards of all the players, and examine them for sketchy play.

But the anonymity makes it impossible to identify a player in multiple sessions to see if their play is peculiarly consistent like a bot's would be.

So it will be extremely difficult to prove if there are bots on Ignition and impossible to prove that there are not.

If the security team gives a ****, maybe they have the knowhow to look out for these things and other telltale signs of bot play. But the word on the street is that most grey market sites are far more concerned about money laundering than they are about botting or other cheating going on.

My own opinion about bots is that they make the games worse in pretty much the same way that multitabling regs with tracking software and HUDs make the game worse, and (except for never tilting) worrying about one and not the other seems a bit off to me. The fun players will lose to both and lose in just about the same degree.

It's collusion cheaters or superusers that pose a greater threat to the underlying integrity of the game. Collusion is as easy as having a phone call going with a buddy while playing.
AlanBostick is offline   Reply With Quote
Old 09-06-2020, 01:01 PM   #308
DonJuan
old hand
 
DonJuan's Avatar
 
Join Date: Jul 2006
Location: Bored of waiting
Posts: 1,440
Re: 2020 HI content

“Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.“
Warren Buffet

“There is nothing so useless as doing efficiently that which should not be done at all.“
Peter Drucker
DonJuan is offline   Reply With Quote
Old 09-07-2020, 12:28 PM   #309
TheToastrOven
journeyman
 
Join Date: Feb 2016
Posts: 264
Re: 2020 HI content

Hi all,

Since this seems to be the only active thread--
I was wondering if there are any decent limit training courses/materials out there. I've heard tales of old cardrunners and deuces cracked materials, but there doesn't seem to be much of anything newer available.

I've been playing a lot of live 20/40 lately, and the games have been fun and really soft. Even so, I'm finding myself just clicking buttons in way too many spots.
TheToastrOven is offline   Reply With Quote
Old 09-07-2020, 12:55 PM   #310
___1___
old hand
 
Join Date: Dec 2004
Location: __________
Posts: 1,552
Re: 2020 HI content

Quote:
Originally Posted by DonJuan View Post
“Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.“
Warren Buffet

“There is nothing so useless as doing efficiently that which should not be done at all.“
Peter Drucker
“Just because something bears the aspect of the inevitable one should not, therefore, go along willingly with it.”

―Philip K. Dick, The Transmigration of Timothy Archer
___1___ is offline   Reply With Quote
Old 09-07-2020, 08:59 PM   #311
DonJuan
old hand
 
DonJuan's Avatar
 
Join Date: Jul 2006
Location: Bored of waiting
Posts: 1,440
Re: 2020 HI content

I guess the quote fit perfectly for a dying game by a science fiction author
DonJuan is offline   Reply With Quote
Old 09-08-2020, 02:03 PM   #312
AlanBostick
Carpal \'Tunnel
 
AlanBostick's Avatar
 
Join Date: Sep 2002
Location: Old Man Red Bull (he/him)
Posts: 10,477
Re: 2020 HI content

Yabbut The Transmigration of Timothy Archer isn't science fiction; it's a mainstream book about the life and death of a thinly disguised Bishop Pike.

It was the final novel that Dick finished before his death, published posthumously, so in that sense it is just as apt.

(There are other works that were published later, but they had been written earlier and were sitting in his filing cabinet or a trunk or something.)
AlanBostick is offline   Reply With Quote
Old 09-08-2020, 09:29 PM   #313
Abbaddabba
Pooh-Bah
 
Join Date: Jul 2005
Location: twitch.tv/abbaddabba
Posts: 5,633
Re: 2020 HI content

Quote:
Originally Posted by AlanBostick View Post
Doug Polk offers four indicators of possible bot play:


1. The player takes about the same amount of time to act every hand.
2. The player doesn’t answer to moderator in the chat or to an alert.
3. The player plays for unreasonable amounts of time.
4. The player plays an unreasonable amount of tables.

Unfortunately, only the first indicator is something that a player on Ignition could be in a position to observe. There is no chat on Ignition (there may well be on Bodog, but I can't observe it from Ignition). There is no way to tell how long a player has been playing, unless you, too, have been playing with them for a long time at the same table. The software limits cash tables to four, so a single account can play in only four cash games at a time -- but I understand that tournament tables are not limited in this way. In any event, players have no way to determine how many tables another player is playing.

Ignition does have the feature that (twenty-four hours after the player leaves the cash game or after a tournament ends) they can download all the hand histories, complete with revealed hole cards of all the players, and examine them for sketchy play.

But the anonymity makes it impossible to identify a player in multiple sessions to see if their play is peculiarly consistent like a bot's would be.

So it will be extremely difficult to prove if there are bots on Ignition and impossible to prove that there are not.

If the security team gives a ****, maybe they have the knowhow to look out for these things and other telltale signs of bot play. But the word on the street is that most grey market sites are far more concerned about money laundering than they are about botting or other cheating going on.

My own opinion about bots is that they make the games worse in pretty much the same way that multitabling regs with tracking software and HUDs make the game worse, and (except for never tilting) worrying about one and not the other seems a bit off to me. The fun players will lose to both and lose in just about the same degree.

It's collusion cheaters or superusers that pose a greater threat to the underlying integrity of the game. Collusion is as easy as having a phone call going with a buddy while playing.
imagine how simple it is to get a program to delay a click by random numbers relative to getting the whole operation on the ground. imagine how easy it is for the program to send the operator an alert any time the chat mentions the accounts name (with several variations) so they could step in to interact. imagine how easy it is to just not have the account for ridiculous hours or playing ridiculous hours.

the best solution by far is having software that distorts the graphics periodically to mess with whatever screen scrape tech they have. if you need someone sitting there manually clicking the buttons then it's really not much different from actually playing.


but there are important differences as to why this is worse than adding another strong reg to the playing pool in that strong regs are generally only willing to play when the profit ceiling is above a certain point where bots will keep on chuggin' as long as the expected profit is > 0.

a recs suckiness is often so marginal that, even if it may have been very worthwhile to play them hu, it's just not worth playing with them at a full table with other strong regs, especially at 10/20 and below. not that it's unprofitable (though in theory it could be), but many will just play other game types or lower stake tables with weaker competition instead. and when regs sit out and focus on other tables it increases the profitability of the ones who remain. the bot otoh will continue to dilute the profit ceiling until there is none to be had.

it's like the difference between wages being pushed down from competition and wages being pushed down from automation.
Abbaddabba is offline   Reply With Quote
Old Yesterday, 10:58 AM   #314
DougL
Too helpful for this post
 
DougL's Avatar
 
Join Date: Sep 2002
Location: Boulder, CO
Posts: 21,753
Re: 2020 HI content

Wouldn't the card sharing benefit of bots in PLO would make it the game of choice for cheats? Knowing that the other bots at the table had folded the A and K of the flush allows the bot in the hand to make insanely profitable plays. The huge variance in the game would seem to hide ridiculous runs.

Quote:
the best solution by far is having software that distorts the graphics periodically to mess with whatever screen scrape tech they have. if you need someone sitting there manually clicking the buttons then it's really not much different from actually playing.
If there were still large $/hour in poker games, the price of a low cost human to interface for a bot would seem to be a small cost to crushing mid/high stakes. If you can hire miners for video games...
DougL is offline   Reply With Quote

Reply
      

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Forum Jump


All times are GMT -4. The time now is 02:22 PM.


Powered by vBulletin®
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
Copyright © 2008-2020, Two Plus Two Interactive
 
 
Poker Players - Streaming Live Online