21/17/3.0/22

[keikiwai]

i think even the "unexploitable" strategy is exploitable if you KNOW someone is using it, and you know what it is

it is exploitable by switching gears faster than the "unexploitable" can adjust.... so you start of at 30/20 and the moment the "unexploitable" notices you're 30/20 you tighten up, this will make him continue to play against you as if you're a 30/20 but you're not... or obv. start tight and loosen up

so, it becomes a game of who is more in tune with the flow of the game

[Unknown Soldier]

when you say uneploitable strategy do you mean one a bot could do?

[iSTRONG]

[b] anyone who does not understand this and why it's important should ask themselves whether they think it would be more profitable to play vs. a table full of 21/17/3.0's or a table full of 22/17/3.0's.

If you think you'd prefer playing the 21/17/3.0's, it means that you believe that 22/17/3.0 is closer to unexploitable play than 21/17/3.0
By thinking in this way, you should be able to figure out what you think GTOS stats look like.

[iSTRONG]

[ QUOTE ]
i think even the "unexploitable" strategy is exploitable if you KNOW someone is using it, and you know what it is

it is exploitable by switching gears faster than the "unexploitable" can adjust.... so you start of at 30/20 and the moment the "unexploitable" notices you're 30/20 you tighten up, this will make him continue to play against you as if you're a 30/20 but you're not... or obv. start tight and loosen up

so, it becomes a game of who is more in tune with the flow of the game

[/ QUOTE ]

Love it. & i totally agree. This is why my 21/17/3.0 is only an average. I can see that I'm not the only ssnl'er who's thought a lot about this.

Edit: sorry i misread your first sentence as "i think even the "unexploitable" strategy is UNexploitable even if YOUR OPPONENT KNOWS YOU are using it. (with which i agree

[bilbo-san]

[ QUOTE ]
i think even the "unexploitable" strategy is exploitable if you KNOW someone is using it, and you know what it is

it is exploitable by switching gears faster than the "unexploitable" can adjust.... so you start of at 30/20 and the moment the "unexploitable" notices you're 30/20 you tighten up, this will make him continue to play against you as if you're a 30/20 but you're not... or obv. start tight and loosen up

so, it becomes a game of who is more in tune with the flow of the game

[/ QUOTE ]

No. This isn't what GTOS means.

We can argue about whether a GTOS actually exists for poker, but if you concede the point that there actually is one, then there is, by definition, no way to exploit it, even if you know what he's doing.

The Roshambo example demonstrates this. Assume we are playing roshambo, and I pick my throws randomly. And you KNOW that I pick my throws randomly. You cannot exploit this strategy.

The S-C hand rankings in NHLETAP are another example. Even if your opponent knows what you are doing, he can't exploit your strategy.

NOTE: This does NOT mean that the GTOS is the most profitable strategy. It means that the GTOS can't be UNprofitable. There's a big difference.

[ikestoys]

iStrong, i've seen you claim that 21/17/3 is unexploitable, and even define it as such... BUT I haven't seen a solid reasoning why.

[cbloom]

Yeah, bilbo is right on.

This is a very rough definition of the GTOS :

If an opponent is playing the GTOS then all strategies are -EV against him except the GTOS itself (which is of course 0 EV).

In some cases the GTOS can be found by a simple iteration :

start with a strategy S0 which you believe is close to optimal
find a strategy S1 which is +EV against S0
find a strategy Sn which is +EV against S(n-1)
eventually the iteration ends, no better strategy can be found,
that's the GTOS

(this only converges right in very simple games, in complex games better methods are needed)

It's possible the GTOS doesn't exist for NLHE, but it probably does since it's been found for lots of simplified versions of poker.

But as has already said it's not necessarilly the most +EV way to play against bad opponents. eg. the GTOS won't value bet really light because it will be afraid of bluff-check-raises, it will check behind more. It will also pay off nits too much; it will still be +EV against nits because it's gauranteed to be cuz it's balanced - eg. it will steal enough pots from them to make up for paying them off too much.

[iSTRONG]

[ QUOTE ]
iStrong, i've seen you claim that 21/17/3 is unexploitable, and even define it as such... BUT I haven't seen a solid reasoning why.

[/ QUOTE ]

I don't think there is a reasoning behind it. I just find it real hard on average to make money of 21/17/3 type players. If i were to find myself on a table of 21/17/3 players (who are as skilled as me), I'd probably play 21/17/3 myself and expect not make or lose much in the long run.

[delta k]

whatever PA plays is unexploitable. so ya, those stats (lol stats mean nothing end lol)

[lucky_mf]

My initial reactions:
- it is somewhat silly to refer to a stat profile as a strategy in the game theoretic sense when the game is multiple iteration of a best 5 in 7-9 card poker hands. As orange pointed out a 3 variable stat profile which is heavily weighted toward pre-flop play may not have much do with how a player plays post-flop when most of the $ goes in.
- If we are going to start making references to game theoretic concepts we ought to be clear on defining the game, strategy space, ect.
- I'm not sure that we get any mileage out of trying to apply game theory to poker because defining the game is difficult and different for every player (whats the time frame?, how many players are there?, who are the players?, ect) and because the strategy space is large (i.e. it is more than a stat profile).
- At any given time every player is doing something that is exploitable in some way. The trick is to figure out what these tendencies are and figure out what adjustments to make before the other player adjust to your adjustments (if said player is even capable of making these adjustments).

Lucky