Thanks guys, yeah I feel like I have hit the bad end of the deck pretty hard so far. However, I feel like I was not playing as well as I should have been on parts of Day 2 and 3 (Strategy wise). However I feel like today I was playing very well and I'm happy about that despite the result.

claudico always has the super nuts when xr allin the river like mac5/insta, no need to tank, programmers should avoid this timing tell somehow.

The programmers are reading this don't give away the bot tells lol

Lol really enjoying this thread/challenge! Go team human!!

Nice fold with the QJ vs the bot's king high flush. I have a question for you though. In the video while the hand is taking place you are speaking with dong and you mention to him while facing the river c/shove that you have no clubs in your range. You raised pre, bet/called the oesd on the 2flush flop, when the flush hits the bot checks to you and you bet 2/3rds, blank ace river the bot checks to you and you make a solid river bet. I would think you are uncapped here. Maybe you play certain cc hands differently but really NO spades with this line?

Unless I'm reading the board wrong in the fuzzy video on my phone and the flush came in on the river. Or you were just saying you didn't have a club in your actual hand.

Keep it up tho, u crushed on the other side of doug's action

nice challenge, enyoing it aswell. At some point Polk was in bot mode himself when catching up some hands, really grinding the machine down. if it wouldve continued the real bot wouldve start crying. But then the professor stepped in again to snap him out of it and he lost like 20k and won back 10 of it in the end.

I cant really comment if his play actually deteriorated right there(im not even a decent player), but there seemed to be some correlation. So yeah, dont let that proffesor play the mindgames the bot himself is incapable of

gl to ALL

Are the distributions of who gets what hands static? E.g. does Jason get all the hands of Doug's bot? Or is it being mixed up during the challenge?

If so, who mirrors who? I'd guess Doug - Jason and Bjorn - BDDK

mixed, you can see who teamed up with who by the background on the stream. the casino team gets the opposite deal of the casting coach team.

if 3 players of the team human can manage to win more than 2 buyins the same session is super impressive.

Wondering what is your and the group's current assessment of the bot. From purely an non-hu player's perspective, I think -14bb/100 is pretty impressive vs the end game bosses. I'm hoping humanity opens the floodgates and removes any perception that bots are close to catching up in NLHE.

Are you saying -14bb/100 is impressive for the bot ? HU bumhunters make in the 20-30bb/100 WR playing only vs big fish, so a -14bb/100 is very poor imo. A random 50z reg on stars would probably not lose as much as -14bb/100 vs top HU players.

this is rakeless w/r so Im fairly sure that a random 50z reg would get crushed for worse than -14bb/100

1. They are bumhunters so they are nowhere near the top. A top player would make much more against the same opponents.

2. They rarely play long sessions where they can really hone in and exploit all tendencies maximally. Over a few thousand hands they would increase their winrate, even allowing for a small amount of adjustment from the fish.

3. They might not have the incentive to try to maximize winrate/hand. Obviously you should play every hand to maximize EV, if you have unlimited amount of action. But bumhunters don't have that. It's more important to maximize EV from every single fish, even if it means skinning them a little slower. E.g risk less at the start to minimize the chance of hit and runs or use whatever tricks bumhunters use to increase the chance of the fish reloading.

What you have here is that after the bots strategy is laid out and analyzed by the best players in the world, they can only find 15bb/100 worth of leaks in it. Maybe slightly more as they understand the bot better. I doubt many players could match that.

Serious question: Is the bot supposed to be "learning" how to play poker better overall from playing these humans or how to play better against each of these individual opponents or an amalgam of these four opponents?

Doug beat sauce for 20bb/100 in the challenge, this bot only loses with 14bb/100 at this moment against 4 top hu players, it might close the gap as the days go on. That laid out, it would be a pretty scary bot for most people.

^^ I don't think anyone seriously thinks that Doug was 14bb/100 better than Sauce.

Are you joking?
1. Do you think 15k hands one way is of any significance compared to 80 k hands played mirrored?
2. Do you think Team Human plays anywhere close to optimal vs the bot had they known it's exact strategy?

Just face it. The bot is great. It can play poker. It can't play at higher competitive levels, but it can play well. It loses vs top human players who play w/o HUDs and who is at times strucken by fatigue in a noisy, unfamiliar environment. The maximum the bot can lose is 75bb/100, which is if it folds every hand. Highly intelligent researchers and some of the brightest minds available has managed to cut this loss rate down to 15bb/100. Factor in how much a GTO bot would kill Doug for and just realize that it's a pretty long way to go before it comes close to GTO, especially since it gets exponentially harder for every step closer it gets.

Is it an impressive feat? Yes. Do we have to fear anything for the coming year or two? No.

I'm all in on under 14bb/100.

He might not have been 14bb/100 better, but it doesn't seem outrageous to think that could have been. There was a RIO video where Sauce mentioned that Doug told him that his limping strategy in that match was really bad and Sauce has since abandoned it as far as i know.

Yeah, but Doug and Bjorn are beating it for over 60bb/100.

But Doug has flopped a set every 5th hand and Jason is getting crushed while playing the same hands from the opposing perspective. So while he's still beating it for a large winrate, his true winrate isn't anywhere close to 60bb/100

75bb/100 is not the maximum possible loss ... if I call my stack off with the nut low, 200bb deep, I just lost 200bb/1

But I agree with you: I personally think the bot seems not as bad as many ppl in here would like it to be, given that it is facing some of the best "brains" out there.

[x] Plays Microstakes
[x] Makes assumption about the bot not being great
[ ] Knows what it takes to win
[ ] Has any idea about winrates

This bot is terrible compared to current online midstakes bots. It already does very well against doug who is a maniacal bot, an asian half-bot crusher and bigdonk kim whom is still down after 5k hands.

Bots > most good humans.

I would bet decent money on you playing one of the current bots HU and losing a lot.

I think I was saying that about another hand and realized I was wrong.

The bot plays exceptionally better than Tartanian7 which they won the bot competition with back in June. I'm pretty impressed with how it plays, and expect this will become even better over time.

http://en.wikipedia.org/wiki/Ad_hominem

[ ] Will discuss what I've achieved in poker with you.
[x] Will discuss the bot, how it's doing, strategies, thoughts about future of poker etc.