It's a statistical measure of volatility. The higher standard deviation something has the swingier it is. Low standard dev means all data points are very close to eachother, high the opposite.

You can probably google up some good explanations if you want to understand it more. It's about how scattered data points are in a set.
5 4 5 6 4 5 5 3 5 =low standard dev
18 2 7 9 6 3 7 1 12 =high standard dev

If the standard dev is low, then a lower winrate can be statistically significant while the same winrate would not be at a higher standard dev. Imagine it yourself: A super swingy poker game will see big swings without it being certain that one person has an edge. One player winning 10 buyins in a super swingy deep stacked PLO game won't tell you much about who is the better player, but if they are playing limit hold'em then winning 1000bb probably is enough to determine who's the best of the two. What exactly 180bb/100 means in a poker sense is: not much. It's normal in HU NLHE. But the number is needed to figure out if winning xbb/100 over y hands is statistically significant or not.

https://pokershares.com/pokershares#...110-113-114-0,

I would bet a math expert/ computergeek, with a moderate feel for poker would have a better chance of winning, then your standard poker pro.

In layman's terms, if I'm not mistaken, 95% of the time after 100 hands you will be within 360bb/3.6 buy-ins of where you started in either direction assuming no winrate.

Also, I'll bet 2.5:1 on the humans up to 25k. So for example my 25k your 10k.

Am laying 2.5:1 on humans up to 3k€ (my 3k to your 1.2k) ,also willing to escrow ,pm if interested

looks like it starts january 11-

https://www.riverscasino.com/pittsburgh/BrainsVsAI/

pretty exciting. really hope humans win, but im guessing the bot is going to be several orders of magnitude greater than claudico so it should be rather close

thinking of betting on this, who's the fourth human, donger, dougiedan, premiumwhey, _____? Sonny_Black?

Pretty exciting; 4 nerds playing poker against a computer is exciting to you? This is what poker has come to?

it's exciting/interesting for a lot of pretty obvious reasons

fortheswarm

I'll take the AI at 2.5 to 1 from anyone in LA who is a reputable high stakes player/scenester or someone who can have a reputable player act as proxy to pay out the bet. escrow preferred unless we mutually deem it unwarranted. outright winner (no statistical ties or reductions of variance, just raw results). willing to risk up to 15k.

aiprop2017@gmail.com to email an offer (min offer 5k). this is ty from the gardens.

So according to current HU rankings, donger is close to top 5 and the other 3 guys somewhere in the #10-20 range?

Does the empire still play? At least on stars? Have not seen swarm, donger, premiumwhey or dougiedan play for quite some time

HU is so dead that trying to make rankings at this point is meaningless.

I'll book action on AI, 1:2.6

up to 20K USD
my 20K USD vs your 52K

PM skype details if interested

amount of ppl wanting to bet on AI seems interestingly high?

i would never bet on this without having more info than others on players, AI etc.
Im assuming the ppl wanting to bet big, have that info to some extent.

just food for thought

Those betting, make sure you decide what a 'win' means...

pretty sure everyones on the same boat now that

win = having more chips at the end

Action still available and I'm undercutting this guy fwiw. Can provide numerous references. Will escrow with someone known and reputable (Limon comes to mind as a famous on 2+2 guy).

bit of a stretch imo... these "super bots" are highly sophisticated and developed over years. if you can afford to run such software (including the millions of core hours for 'learning') and then try to cheat at online poker, you prolly did something wrong.

tl;dr: the real threat are the way less sophisticated random bots you can buy and tweak.

Willing to bet on computer, my 1 to your 2.5, up to 4k.
Need to escrow.
Bitcoin, Stars or Skrill in Euros is best.

Please PM me Skype details if interested.

hopefully les pulls his weight this time

How many buy-ins does either side have to be up after 120k hands in order for a formal win to be declared? Does it depend on the data set from the match itself? (and therefore can only be estimated based on the previous one?)

Breaking: HUNL is "solved": https://arxiv.org/pdf/1701.01724v1.pdf. I guess if the CMU AI wins, they should arrange a match against DeepStack.

@ShallowMind, a win is declared if one team is one std deviation above zero which corresponds to a statistical 95% confidence interval. Std dev can only be figured once the match is over, but was about 180bb/100 hands in last match. Maybe on the high side because many players had aggressive style to avoid the AI tanking on the river.
If my math is right a team should win on more than about 6 buy ins to be significant?