Thoughts on using randomizers for a mixed strategy - Small Stakes Pot Limit Omaha

Two Plus Two Forums Poker Strategy Omaha

Thoughts on using randomizers for a mixed strategy

Post Reply Subscribe

...

12-21-2018 , 03:11 PM

HUMBLE.

veteran

Join Date: Aug 2017 Posts: 2,010

A randomizer just chooses a number lets say from 1-100.

The solver (such as Monker or PIO) tells us to check 25% of the time and bet 75% of the time.

We run the randomizer and the number lands on 1-25. We check.
We run the randomizer again and the number lands on 26-100. We bet.

Sounds good? Or is this misapplied? What is wrong with this thinking?

Quote

12-21-2018 , 04:13 PM

VerdantDevil

grinder

Join Date: Apr 2013 Posts: 598

Scenario 1: Your opponent is an ultra nit. With your hand, you are 'supposed' to bet 75% of the time. In this scenario, the randomizer tells you to check. You check.

Result: You fail to easily steal a pot against a nit

Scenario 2: Your opponent is a tilting fish. You flop the world. With your hand, you are 'supposed' to bet 75% of the time. In this scenario, the randomizer tells you to check. You check.

Result: You fail to capture value against a villain who is likely calling you down.

Solvers only work assuming your opponents all play GTO. Do they in reality? Mine don't.

Quote

12-21-2018 , 04:31 PM

+EVillain

grinder

Join Date: Jun 2018 Posts: 569

In PLO the best way to randomize between mixed strategic options is based on side cards, unlike NL each hand has so many characteristics that we (and solvers) mostly use those to pick between options. So the best way to mix is to let the secondary characteristics in a hand guide our decision one way or another.

For example if a sim advocates mixing 50-50, if you narrow the syntax of your search on the hand categories you will start to see trends in the way the solver is advocating you mix based on blockers, additional EQ, etc. flush, pair, straight, and set blockers are a huge part of controlling our frequencies.

Quote

12-22-2018 , 04:32 PM

pucmo

adept

Join Date: Mar 2016 Posts: 1,187

The opponent, nut blockers (when only the nuts matter), the random (finish off misery and be balanced for sure or as exploitive), reverse, the technical (less meaningful) "side" cards.

Quote

12-24-2018 , 08:51 AM

HUMBLE.

veteran

Join Date: Aug 2017 Posts: 2,010

Quote:

Originally Posted by VerdantDevil

Scenario 1: Your opponent is an ultra nit. With your hand, you are 'supposed' to bet 75% of the time. In this scenario, the randomizer tells you to check. You check.

Result: You fail to easily steal a pot against a nit

Scenario 2: Your opponent is a tilting fish. You flop the world. With your hand, you are 'supposed' to bet 75% of the time. In this scenario, the randomizer tells you to check. You check.

Result: You fail to capture value against a villain who is likely calling you down.

Solvers only work assuming your opponents all play GTO. Do they in reality? Mine don't.

Playing to maximize exploits leaves you open to counter exploitation. Players typically adjust to this by countering the counter exploit and so on in a dangerous cycle.

Quote

12-24-2018 , 09:02 AM

HUMBLE.

veteran

Join Date: Aug 2017 Posts: 2,010

Quote:

Originally Posted by +EVillain

Let's take blockers for example. We see data from the solver check raising hands that contain top pair with a high frequency. From this we might conclude that the solver does this because we block top set. We are giving reasons to data and I find that confusing because the solver does not "think" in terms of reasons that we as humans need be able to take action.

Last edited by HUMBLE.; 12-24-2018 at 09:30 AM.

Quote

12-24-2018 , 09:53 AM

SeaKing

Pooh-Bah

Join Date: Feb 2010 Posts: 5,387

You have the wrong state of mind

Quote

12-24-2018 , 10:44 AM

HUMBLE.

veteran

Join Date: Aug 2017 Posts: 2,010

Quote

12-25-2018 , 10:26 PM

+EVillain

grinder

Join Date: Jun 2018 Posts: 569

Quote:

Originally Posted by HUMBLE.

Yeah it’s true, sometimes we have to extrapolate or whatever, but our best guess is usually right. The bottom line is that these solutions are accurate, so even if we somehow attribute the wrong meaning to some aspect of the solution it doesn’t really matter

One thing that really interested me about PLO solver when first exposed is that they will run some small amount of bluffs without blockers (Pure air), depending on the situation. The solver is only concerned with realizing EV, it doesn’t need to justify its bluffs based on blockers or whatever (like a human usually does)

Last edited by +EVillain; 12-25-2018 at 10:33 PM.

Quote

Post Reply Subscribe

...