Out Now
×

Humans vs Algorithms

Posted by

You’re watching:

Humans vs Algorithms

user avatar

Nuno Alvarez

Elite Pro

Video Player is loading.
Current Time 0:00
Loaded: 0%
Duration -:-
Remaining Time 0:00
  • descriptions off, selected

Resume Video

Start from Beginning

Watch Video

Replay Video

10

You’re watching:

Humans vs Algorithms

user avatar

Nuno Alvarez

POSTED Feb 24, 2023

Nuno Alvarez shares his thoughts on the differences between humans and algorithms approaching poker. He discusses why solvers can enlighten our logical understanding of the game in almost endless ways while still forcing us to fall back on logic every time we face real opponents at the tables.

15 Comments

Loading 15 Comments...

mx404 2 years, 1 month ago

Hey Nuno, cool video, I enjoy it and find it helpful.

I really like this type of “big picture” theory base video. And I do enjoy your storytelling, making those complex concepts easy to digest.

Wish you can make more in the future, cheers!

Nuno Alvarez 2 years, 1 month ago

Thank you very much mx!

Seeing how well received this one has been I'd definitely make more theory videos in the future. Stay tuned :)

matlittle 2 years, 1 month ago

Good video - I agree with the points regarding how humans will differ from solvers!
With regards to the inelasticity of calling ranges - how far would you go to exploit this against a human? Seems like our bluffs are not indifferent between different bet sizes on the river, and would prefer to use the smallest size we use on the river. e.g. if we are mixing between 75% and 150% bet sizes on the river, they would prefer the 75%. How far would you take this adjustment in game - would you put all of your bluffs in the smaller sizing, or just a few more than a solver would suggest?

Nuno Alvarez 2 years, 1 month ago

Thanks mat!

Seems like our bluffs are not indifferent between different bet sizes on the river, and would prefer to use the smallest size we use on the river.

Can you elaborate a little bit more on this? Why would our bluffs prefer to use the smallest size we use on the river?

matlittle 2 years, 1 month ago

Sure, so my logic is that if people are inelastic to how they defend with regards to what bet size you use, then you are incentivised to use a big sizing with strong value hands and a small sizing with bluffs. For example if you 3 barrel a board and think that on the river your opponent will call almost any top pair vs a bet regardless if you bet 75% or 150%. If they are supposed to fold plenty of top pairs vs the 150% at equilibrium but are not elastic enough to the size of the bet, then bluffing for the 150% size instead of the 75% size will be lower in EV (and probably -EV if they are really inelastic).

Nuno Alvarez 2 years, 1 month ago

I see what you meant now.

It all comes down to how inelastic you expect your opponents to be of course. I do agree with your logic that if they are going to bluffcatch a similar range no matter which size we use then we should obviously always and only bluff for the smaller size and always and only valuebet for the bigger one. However, this seems like an oversimplification that will mostly apply only if the gap between the ''smaller size'' and the ''bigger one'' is quite small. If we use a blockbet (25%) and a PSB (100%) as our example, then even the weakest players would be capable of understanding that they need to defend much more when they face the smaller bet than when they face the bigger one.

What I meant in the video when I mentioned this topic is that humans are significantly worse than the machine when it comes to understanding how much they have to defend regarding the betsize they're facing. Instead, we tend to focus almost exclusively in how good our combo is rather than how good it needs to be given the betsize we're facing but, as I said, this whole idea will be applicable mostly when the gap between sizes is quite big.

matlittle 2 years, 1 month ago

Do you have a general heuristic for where the non-natural bluffs are pulled from by the solver? Is it mainly just based on number of overcards to villain's 1-pair calling region?

Similarly do you know much about the very unnatural bluffs - i.e. close to 0 equity bluffs? Sometimes for example I think you see the solver overbet with hands like 54s or 98s on the turn on AXXX with no chance to improve/no overcards to even 2nd pair. Presumably they block BB's top pairs but I am never sure which scenarios in which to use a hand that is almost drawing dead for a turn bluff.

Nuno Alvarez 2 years, 1 month ago

Great question.

I think that overcards are certainly a big part of our non-natural bluffing hand choices since trying to have equity vs the calling range when bluffing before the river is a very common trend on PIO.
However we do sometimes tend to use those close to 0 equity bluffs that you mention and I think that the main purpose of those is to cover certain runouts in which we will end up lacking bluffs if we don't find those hands at any frequency on the previous streets. AFAIK, at least in HUNL, we almost always need to find some of those hands OTT and the qualities that we look for in those are usually unblockers to the folding range (when ranges are wide) or blockers to the continue range (when ranges are narrow)

I actually already recorded a video about something really close to this topic and it is just coming next so stay tuned for that!

matlittle 2 years, 1 month ago

the qualities that we look for in those are usually unblockers to the folding range (when ranges are wide) or blockers to the continue range (when ranges are narrow)

Cool, I will try to look out for those patterns!

I actually already recorded a video about something really close to this topic and it is just coming next so stay tuned for that!

Looking forward to it!

Be the first to add a comment

You must upgrade your account to leave a comment.

Runitonce.com uses cookies to give you the best experience. Learn more about our Cookie Policy