So i played this bot last night for around 400 hands , i heard the bot can not adapt to your play style and i feel like he has many leaks bluffing way too much boards where i simply never fold and where nobody ever fold, also he value thin and its veryeasy to check call him flop raise turn and check raise again on the river he always bet his missed draw or pairs in this spot , but i had few stranges hands like this

https://gyazo.com/df228c1f6aff8d737a81e41df25047

i checked call flop turn he bet i check raise he call river i check he check back with 2 pair in position this seems super suspicious to me .Your guys got any tought on this ? Is this bot rigged or can he simply adapt to your playstyles ?

https://gyazo.com/df228c1f6aff8d737a81e41df2504746 this link sorry

I've not played with cepheus but it's my understanding that it's frequency based.

So, given prior action, it may bet this river 73% of the time or something like that.

bot block all hands that you would c/r and then check riv that it beat 5x. the hands that you would also c/r turn consist of 74s, two pair,33. I think this is one of the hand the ev of bet/folding is worst then check so it choose the option that max. I mean your play pretty amateur maybe can be exploit against human but you playing a against a bot that "rational"

Yes.

Also, it's not adapting to anything.

skilled, i think you have a misunderstanding of what cepheus does. it's not built to adapt to anything. it's built to play an unexploitable game, or be as close to gto as possible. the reason it value bets thinly in a given spot would be due to it's hand, pot size, and it's game plan against another gto player in that spot. it's an extremely well built bot.

you didn't post how you did against it over those 400 hands?

it may seem like it bets too frequently on some rivers where nobody's folding b/c it HAS TO select hands from it's range in that spot to bluff with. so if it got to the river by k/r'ing the flop with a draw and it misses, even if it checks the turn (which it'll do sometimes), it'll include enough hands to bluff exactly betsize/potsize% of the time on the river.

I agree, cepheus is a fish.

UpHillBothWays, i did extremly bad gainst him i lost around 1 k chips but i also ran extremly bad, i did bet money against my friend 800 dollars that i will finish up against this botafter beating him a bit in training mode earlier , but the sample of hands is just too short a lot of variance .Idid this cause i find too many leaks in his game and cause not adapting to my play style i just been check raising so many boards in bluff against him since he cant adapt and dont playback at me .

Let us know if you want more action on that bet

J Lot

yea it's pretty clear you don't really know what's going on in terms of gto strategy. as jlot said, i bet you can get a LOT of action on that bet of yours whenever you want. please feel free to post here if you'd ever like to arrange action for you vs. cepheus.

the main giveaway is that you keep thinking that cepheus should be adapting to your play and that it has to play back at you. it DOES NOT AT ALL need to adapt to your play. instead, it was designed to play gto strategy (or as close as anything can get to the purest form of it. it does collapse trees on the outside margin, but that makes it usable in terms of the set of calculations it needs to do vs. taking into account every single possible eventuality which would make it impossible to utilize). that way, YOU cannot exploit IT.

it doesn't give 2 sh*ts how you play. all it knows is that it's chosen its buckets and associated actions based on the pot size and it's overall strategy. so yes, it'll lose money by not super taking advantage of a maniac, but it also means it is extremely tough to play against.

anyways, hopefully you need some more action soon. let us know and i'm sure you'll get booked instantly.

Yes please

+1

Basically, if you go to a heavy x/r strategy, you're over-bluffing. The bot makes money on this mistake by calling like it would vs. a perfect opponent. It might make more money vs. you by calling even more often, but it still profits playing as it does. As UpHillBothWays said, it doesn't have to adapt to you.
the belief of people better at HUHU than I am, is that the bot is really good at HU LHE. It is certainly true that there is a lot of variance in HU poker, however, that only helps you against a strong opponent. Forum nice-guy Howard Beale beat one of these bots for a while and posted about it in the LC thread. Like me, he's not a HUHU pro. A number of pros offered to book action vs. him, and he wisely declined. He did post results over bigger samples -- as expected, the bot crushed.
Yes, to be very good at LHE.

i just realized that we may need to take a step even further back here

skilled: gto means "Game Theory Optimal"

this is a mathematical subject area initiated by nash and it is designed for deriving optimal strategies in games with thinking/reacting opponents and incomplete information. the purpose of GTO(gto) play is to ensure that no matter what happens, the opponent cannot win by employing an algorithm of any kind whatsoever.

in other words, the result of a gto bot is a bot that successfully implements either: a) a strictly dominant strategy (i.e. one that outperforms all other strategies in expectation, or b) an equilibrium strategy (i.e. one that results in a stalemate in expectation if all expert perfectly playing players play in this fashion)

since a) is basically impossible in poker, and since b) provides all that is needed to generate an expert playing bot, the UofA folks focused on developing a mixed strategy that suuuper closely approximates a nash equilibrium. the result is a bot that plays its strategy so well that it gets close enough to the target nash equilibrium that it is indistinguishable statistically over something like 60million hands. so in effect, it was designed to be unexploitable (i.e. can't lose). THAT'S the purpose of a gto optimal strategy. it's to avoid exploitation no matter how good the opponent is.

hopefully this helps. if you'd like to learn more, you should read how they designed the bot:

http://webdocs.cs.ualberta.ca/~games...ai-cfrplus.pdf

The important takeaway here is that any deviation from gto will result in a loss of value equal to or greater than zero.

The main takeaway here, is that you got hustled for $800, and you should be working on a buy out.

i posted a few hands against cepheus, OP

please feel free to read and chime in

and so far i'm up still over a few hundred hands lol

How do u see your results? Just played first 100

It gives you a running tally on the left.

I was on an old iPad so I don't think that showed up, will try on pc next time thanks :-)

Some interesting development OP think he greatest gifts against regs on stars and start running this mouth. Obvious one thing led to another and we might have an interesting either at 6 max or head up prop.

I think this hand sum it all on what I am dealing with here.

He does split all pocket pairs 22-88 and 67 is a huge part of your turn semi bluff range. Play seems reasonable iyam

yes, 3betting turn A+

Breaking news: someone made DonJuan feel insecure about his pokering again so he challenges them to huhu for rolls and posts a shame hand to prove he has the biggest poker dick

Not so breaking news: everyone one of us has at some point botched a 3b bluff and backed in to a river we had to unfortunately call because our bluff bloated the pot

Wow this thread actually got good!