If there were 4 diamonds on the board then betting the low end of a 4 card straight with no diamond is not a v-bet

well, doug said it was a v-bet

What would be the value of this bot if it were able to work on stars?

^I thought he rated it a 4/5 at one point...maybe that was day 1, before the engineers/whoever made adjustments?

he said it was a 4/5 compared to 25/50 regs

Do you think this experience will improve the ability to identify future bots?

gotcha, ty

No human can copy this bots style effectively. If this bot was playing I believe we could identify it pretty easily

Jason (or any of the guys) how much work are the 4 of you doing on this in between sessions? Can you talk about any "game plan" you guys had going into this?

wow 9/10 is a huge improvement/"victory" for the bot creators. I see some misintereprations are flying around, Doug said after the first day that he rated the bot 4-5 out of 10. not 4 out of 5 which made the professor real butthurt, but now its 9/10 appearantly.

Is team brains really 700k up or is it error?
Last time i checked Jason L. was 200k down.

i think its gonna be updated to about +220K for humans?

pretty insane how fast the bot is improving. its like the hulk getting stronger fast when you hurt it. as a puny human i wouldnt have thought these 'couple of hands' of trillions hands its played already would make such a difference.

the humans are plugging its leaks basically.

im not sure thats how the bot works

http://www.twitch.tv/claudico_extra

is that a displaying error for jason? how the **** did he win 40 buyins since I last checked?

His last session was +30k, don't know exactly what his total was to start of the session, maybe they just added it to and assigned him the positive score, so Jason should be at -170k now(?) Or it's just completely fekked. Anyways, unfortunately no 400k session for him

We have been able to do work for a couple hours most nights but unfortunately have to cut it short sometimes as we are only human and do need some sleep. Can't say anything about game plans we had going into this, sorry.

This is not how it's working

And yeah the scoreboard is wrong, we are like +260 total I think.

They certainly know what their own ranges look like, and which parts of it they need to defend vs shoves, don't worry about that.

Never mind, have you done those daily range reports ? If so, do you care to explain what exactly you are doing and to publish results here, any interesting finds so far ?

Clear valuebet bud, u might find it too thin but it is not a bluff.

bot has an unfair advantage by having the ability of analyze every hand regardless there was or wasnt a showdown, a normal player could pick up this many tells/reads by 100's of tousands of hands this bot has done in 7000 hands/player. i think.

I remember 1 flip so far: JJ v AK

AI team gets all the hands after the session, humans as well, no ? Where is the advantage for the AI team ?

i don't know, its analizing real time as well, not? meanwhile... it can do all the work 24/7 what human brain cannot. there is not enough time between sessions/sleep to really consider the conditions fifty-fifty. 3000 hands is a ton to do the work hand by hand.

The humans obviously disagree with the bot about having a limping range at all. Most likely the bot can have 100% of hands when both limping and 2.5x pre. It may be weighted such that worse hands are limped more often than better ones but just because the bot limped you can't rule a hand out. Notice the bot limp reraises fairly often indicating it does limp premium hands for balance. The humans must disagree with the bot's strategy as they are exploiting it everytime it limps with a 4x raise with any 2 (usually followed by the good ol minbet).

Also, the min leading strategy really has nothing to do with preflop vpip. They call the bb because it is better than folding or 3bet independently of the strategy they choose after seeing the flop fall. I can speculate as to a few of the initial reasons they chose the minlead strategy and why they continue to employ it (besides the fact it seems to be working and possibly 'confusing' the bot). If you were to choose this minlead strategy vs most good human opponents they would quickly come back at you with their maximally exploitative strategy; they would see you are making suboptimal plays and most likely try to punish it as best they can. In position this would consist of raising that minlead a ton. The bot however does not adjust to its opponent in the same way a human does. It chooses the most balanced GTO line it can at all times. A minlead is not too far from a check in terms of ev especially as the pot to minbet ratio increases (minbet is closer to a check in a 4bet pot than a single raised pot). A check is the standard approach to most hands oop without the initiative so it's not a huge Variation from the human player's normal strategy. The bot may have trouble with this size for technical reasons. If the bot treats the bet much differently than a check it could lead to a situation where the bot is flatting minbets in spots where it would be betting larger than min when checked to. In this scenario the human players are giving up very little to reduce the bot's average pot size in position.

The minbet style also could have just been an experiment and the players liked the way the bot responded and find it easier to play/read with these lines and are therefore sticking with it until Claudico gives them reason to do otherwise..

this limping/opening mixed strategy is a thing that just a bot can balance reaaaaly well.