GTO+ question - why is solver choosing lower EV play with substantially higher frequency? - Micro Stakes Pot Limit and No Limit

Two Plus Two Forums Poker Strategy Online No-Limit Hold’em Cash

GTO+ question - why is solver choosing lower EV play with substantially higher frequency?

Post Reply Subscribe

...

03-25-2020 , 12:59 PM

silent_poultry

centurion

Join Date: Jun 2019 Posts: 150

Hello guys, just a quick question, I've been playing with GTO+ solver for quite some time already (still not using it to it's full potential probably) and what I often tend to see when I solve some spots is that solver is choosing a lower EV play with a bigger frequency for some combos. The answer is probably out there, and maybe it's relatively easy but I'd like to understand. I'll rephrase it and show you:

Here is our strategy for KQo on some flop, I gave it an option to bet big, bet 1/3 and check. Solver is not using the small bet at all, yet it shows the EV for this exact combo would be higher if we bet small. What do I miss? Thanks in advance

Quote

03-25-2020 , 01:16 PM

dubakkur2

veteran

Join Date: Jul 2019 Posts: 3,111

can you provide further info?

it all depends on:

1) SRP / 3bet pots, eff stacks
2) Position 1, Position 2 ranges input
3) Board texture -
4) dEV settings

Quote:

KQo on some flop - doesn't help me

Quote

03-25-2020 , 01:28 PM

DooDooPoker

Carpal \'Tunnel

Join Date: Mar 2018 Posts: 19,699

Solvers will never choose a lower EV play. They always take the highest EV line by definition.

You are looking at the wrong EV tab. You need to go into advanced settings and add the tree and look at it that way.

Quote

03-25-2020 , 01:37 PM

silent_poultry

centurion

Join Date: Jun 2019 Posts: 150

@dubakkur2

1)SRP, 60bb effective, it's an MTT spot
2) HJ vs BB
3) 9h6h4c
4)0.25

@DooDooPoker

You mean tree navigation? If I take my mouse over a branch it still shows that check for example OTT (55% in that scenario) has something like 4000 EV in chips, whereas overbetting has 7400EV. I just wonder, why then we check 55% of the time and not overbet more for example

Quote

03-25-2020 , 01:40 PM

Zuko

veteran

Join Date: Nov 2015 Posts: 2,356

if the solver would acutally take that option 0% of the time, it would not display an EV amount for that action.

when the solve starts, every action is picked at the same frequency. while running the solver adjusts and picks some actions more frequently. the solver realised that betting big is higher ev than betting smaller so he is mostly checking or betting big. BUT he is still betting 0.001 Combos small (because the solver cannot solve for 100%, it is only an approximation). because the action is taken so rarely (less than 0.1%) the EV display does not really matter or count... the solver just didnt get to finish his "job"

Quote

03-25-2020 , 01:45 PM

silent_poultry

centurion

Join Date: Jun 2019 Posts: 150

Quote:

Originally Posted by Zuko

I agree and fully understand that but you're talking about a situation where it chooses to almost never select the second option, however what I'm trying to understand is the fact that when we give it 3 options, bet small, bet big, and check and it chooses those 3 options with some frequencies, I can see that lower frequency option has the higher EV in the program. That's what I'm trying to comprehend.

Quote

03-25-2020 , 01:48 PM

dubakkur2

veteran

Join Date: Jul 2019 Posts: 3,111

Quote:

Originally Posted by silent_poultry

if it is not time intensive - can u run with a tighter nash distance? see if the result changes ?? - may not have converged properly

Quote

03-25-2020 , 02:08 PM

Zuko

veteran

Join Date: Nov 2015 Posts: 2,356

no. if the solver would be able to fully solve the sim it would choose to never use the 2nd option. but the solver stopped before it could get to that result. if you would force the solver to pick the 2nd option more often the EV of the 2nd option would be lower. the ev result of option 2 is not represantative when that option is only picked with a 0.000000000000000001% frequency

Quote

03-25-2020 , 02:18 PM

Ojune

Pooh-Bah

Join Date: Feb 2017 Posts: 5,171

solver does not take a lower ev play ever

Quote

03-25-2020 , 02:53 PM

#10

silent_poultry

centurion

Join Date: Jun 2019 Posts: 150

Ok cool so I guess I'm just reading wrong what it shows? Because it is perfectly logical that it takes the highest EV play, but when you have raise or call for example, it shows 4k chips for a raise and 2k for a call in terms of EV. I guess it's not "relative" EV as you can call it, but I still would like to know why is it showing like this

Quote

Post Reply Subscribe

...