When you know nothing about a situation , is it better to say that all possibilities have the same chance to happen than say some possibilities have a higher chance than others? for example is it better to say someone has 50% chance to bluff or vbet instead of saying he has 75% chance to vbet and 25% to bluff?

Lets consider this hand , a villain from unknown stakes comes to my fr table and i know nothing about him , i only know one thing he raise like 12-15% of his hands at mp , everyone have 100bb , he raise 3bb from mp and i call him from the button with my AQo , everyone else fold , flop is: 8-5-2 rainbow. Im obviously ahead in this flop , most of his range are unpaired cards which are below AQo.

Nevertheless he bets the flop for like 60% of the pot , in this situation what is the best mindset i should have? since i know nothing of him do i say that he has equal chance to bluff me and vbet me? is it in theory better to consider this than saying he vbet only instead?

I call his bet... turn is: 2 , he fires a second barrel , now what? nothing changed , if possibilities remain the same then i have to call this and if nothing improves on river then i have to call river too.

Its too easy to see if your range is ahead or not but it doesnt end there , there is a constant battle where one player tries to convince you that he is ahead despite nothing changes and you have to arrange a possiblity to him and react.


So in theory whats the best way to think about these situations with no reads at all against a total random? whats the most optimal way to arrange possibilities on actions? or all arrangements with no reads are equal? therefore folding the flop or calling a triple barrel yields the same ev with no info at all?


Of course none knows the optimal way to play a hand since we dont know gto but nevertheless we have to play these situations in some way.... how do you consider possibilities in these situations? how do you play them out?

Bet if you think you're ahead.

Not sure we'll ever find exact GTO for NL holdem but if we do I don't think GTO will involve a lot of "bluffing" (in the usual sense of the word). It's just ranges vs. ranges. Bet if you've got a good hand.

Call if you've got a good hand.

This is a borderline hand but anyway what do you think about villain's bets here? how % do you give him for vbets/bluffs? its a total readless situation , how do you arrange possibilities for him?

Interested if you would elaborate on your thoughts about what you think traditional bluffing is and why you feel it won't be a part of a GTO solution.

Base the decision on population tendencies.
e.g. If most "known" players in a pool are mostly c-betting with strong made hands (or the opposite - they c-bet with air at a high frequency), then there's a decent probability that an unknown player is doing the same. That said, it might be the case that an unknown player is slightly more likely to be a weaker player, because if he was a reg, you'd be more likely to know him. Personally, I revert to "default" actions until I have a large sample size (and/or notes) on anyone.

I'm not sure what you mean by the part in parentheses, but GTO involves a quite a lot of bluffing. It's hard to make a pair in holdem, and it's impossible to make money against good players if you only bet with strong made hands.

Oh, this spot has little to do with villain being an unknown or not. It's time for you to do some range work.
Start by putting yourself in villain's shoes. Pull up Equilab/Flopzilla and input a "typical" 12-15% range. (It might look like a bit like a tight 6-max UTG range). Now look at that flop. Ask yourself which hands you think you'd bet.
Now ask yourself how they would proceed on various turn cards if button calls on the flop.
Now ask how AQ plays as a call against the range that bets the flop. Does it have much equity against some of the "value hands" (overpairs, sets)? Will villain be one-and-done with his air, or will there be plenty of good barreling cards for him? Will you have a chance to steal the pot on the turn if villain checks? Could AQ even be checked back with SDV? You need to work out how your profit (if any) will come. Will it be from sucking out, or by betting to pick up dead money (possibly with the best hand) or by simply checking down? In short, you need a plan for the hand. If the average EV (which you can only estimate) of all the likely outcomes isn't profitable, just fold directly.
Another way to look at it is with defending frequencies. If you picture your entire range for this spot, you might find that you have many more combos of "better" hands that can continue versus a c-bet (enough, perhaps to prevent villain from auto-profiting with a half pot bet). In which case, folding AQ becomes a lot easier. I haven't studied or simulated this particular spot, but I imagine that AQ is either at the top of your folding range, or bottom of your calling range, so it's close to breakeven whatever you do.

What I meant is that I think a perfect player would use a merged range rather than a polarized range.

By merged, I mean the computer is betting it's top hands most frequently and it's lesser hands less frequently. This doesn't mean that it never ever bets weak hands. It just means that a computer would make pure bluffs less often than a human. And it will bet it's weaker hands less frequently than a human might.

I am not a very good poker player. I'm the first to admit it. But I like the logic behind using merged ranges in theory.

What I mean by a pure bluff is something ridiculous like triple barreling 24 off or the like. This is just an extreme example.

I don't know about all these specific situations. There is a tendency to get bogged down in the details. A computer can approximate solutions much better than a human can. That's all I'm saying.

I don't think a GTO solution for NL is feasible in the near future but I think coming close enough to beat all but the top players is feasible. There will be some bluffing involved but mostly it will just be betting your strong hands imo.

Something did change. Unless he does the same thing 100% of the time you received information from the Villain when he bet the turn.

I like to step back in these or situations and look at it this way.

Assume an ace and a queen are dead (they're in your hand) when you deal 8 other hands to a nine-handed table: 43% of the time someone has AQ/AK/22+. It's not like AQ crushes 8 random hands. It's behind 43% of the time pref-flop.

Some of the the hands that were worse preflop (like A8 suited) have paired and are now ahead of you.

On the other hand if someone rarely value-bets thinly and 3-barrels often they'll have a lot of hands in their range that are worse than yours.

Flopzilla comes in handy to see how someone's range hit a flop.

Post-flop, there is almost no logic to betting "merged" or linear ranges.
Optimal players tend to polarize their ranges as the hand develops.
i.e. They bet their value hands because they want to get called, they bet their weak hands (bluffs) because they want to force folds. They don't bet mid-strength hands that can win at showdown, but can't withstand a raise.
I think this basic theory was in Sklansky's 'ToP', and it's just as relevant today.

Betting mid-strength hands on the river, in some kind of weird range-merge, is one of the most common mistakes that bad players make.


No only do the strongest bots (and players) bluff at higher frequencies than most humans, and with unconventional bet-sizes, they will also turn weak made hands into bluffs (by shove-raising the river, for example), which is something many "solid" players are incapable of doing.

Bonus funny hand played by two of the best NLH bots:


[x] Overbets 2x pot on the connected monotone flop with 4-high, no draw.
[x] Pots the turn that makes a 4-liner.
[x] Checks back the river when it realises villain is unlikely to fold a straight or flush, let alone both.
[ ] Standard.

Good conversation guys. OP - IMO I think the mindset of forcing yourself to allocate a distribution of possibilities and trying to act "accordingly" isn't the best way to approach it. (ie: he has 75% bluffs, 25% value - so I bluff 75% and fold 25%). How the hell do you know which one to choose and with the right frequency?!

I think you have to respond with the optimal action when considering your hand vs. opponents range (absent cbet info/reads). Assume the player is "normal" and cbets much of his range here. As you said you are ahead of his range. However, there isn't much you can rep (it's mostly A high) and he should be aware of this. If you call flop, and brick turn, to him it is basically a green light to cbet turn (which is the same action you'd get vs. over pairs), and if you raise the turn it's still very hard to rep anything, now you've narrowed villain's range somewhat towards hands that actually beat you, and it has become a lot more expensive if you're wrong. Now you are in a difficult decision because you're guessing. I think best action for this hand in situation is to raise the flop cbet. You're not raising for information, you're raising to take the pot against a range you are likely ahead, but on turn/river it's a complete guessing game. This way you can represent strength (or at least show this pot could get expensive), and give opponent a chance to fold. I think you take this action every time with this hand in this exact situation (facing cbet with no info or reads with a hand that is likely ahead but not by far, and you could get better hands, AK, to fold). Also you are protected by having sets (although flatting will happen pretty often) and mid-pairs in your raising range. Also, often by raising the opponent will check the turn with hands that beat you and you can see a free river if you choose. There is no distribution of sometimes call, sometimes fold, sometimes raise. Your action is based on the information you have now.

Calling and inviting an unknown to build a bigger pot is dangerous with this hand. Are you really going to call down 3 streets unimproved with no reads?

I think this perceived distribution of actions is just the result of optimal play, and something observed after the fact. In game the decision is made based on the current information, and it's very consistently made. There is also the measurement problem, where you may think you are distributing actions in one common situation but in truth those situations are differentiated by very subtle differences in information, and shouldn't be grouped (it's not really apples to apples).

Well, I get what you are saying. I have tried both techniques and both have their merits. Maybe I don't incorporate them correctly.

I get similar results either way. I still stand by the linear approach though. I don't think Snowie is perfect yet. I think it is flawed. As for limit holdem that's a different story. But I don't think Snowie is perfect.

Also, I think that good players bluff more often because they have caught on to the idea that people call too infrequently.

I don't think their bluffs are good plays. I think they are getting too many bad folds.

Perhaps there are three solutions: using a linear range, using a polarized range, and perhaps the best...switching between the two randomly....