Decision Trees

#1 July 14th, 2020, 1:57 PM

So I am wondering if anyone here has some advice on a decision making process.

Particularly, if there is some work-flow you follow with each street to determine what actions you might take. I had naively not thought of this until recently purchasing a guide on this very topic specifically for playing no limit hold em.

It seems to make sense in theory, and the system I learned seems reasonable. But I'm wondering what else is out there.

I understand there are a lot of conditions and factors to determining the correct course of action, but there has to be some more info on how to generalize/summarize the process so the player wins more often than not over time.

I would be greatly interested to hear from some more seasoned players on what affects their decision making processes during a hand.

Thanks in advance. Decision Trees

#2 July 14th, 2020, 6:13 PM

Originally Posted by bellicoso

So I am wondering if anyone here has some advice on a decision making process.

Particularly, if there is some work-flow you follow with each street to determine what actions you might take. I had naively not thought of this until recently purchasing a guide on this very topic specifically for playing no limit hold em.

It seems to make sense in theory, and the system I learned seems reasonable. But I'm wondering what else is out there.

I understand there are a lot of conditions and factors to determining the correct course of action, but there has to be some more info on how to generalize/summarize the process so the player wins more often than not over time.

I would be greatly interested to hear from some more seasoned players on what affects their decision making processes during a hand.

Thanks in advance. Decision Trees

I have had similar thoughts and have tried to create a couple decision trees for a book I am dabbling with. What kind of happens is, if you try and create any poker computing program, you must document for yourself the decision logic that your program is going to follow. I created a poker simulator for myself, and it really becomes clear how a computer might work its way through a poker hand. What is unclear is how a human does the same task. A computer can easily take some trail that a human could never do.
First, I doubt that a single tree is possible, without being extremely complex. There are well over 60 variables that could be possible as well as 60+ more questions that could be asked to cover all situations. I ended up with two trees - a preflop decision tree and a postflop decision tree. The reason for this is mainly because once we know who is actually in the hand, we now have a lot more information and questions to ask.
If the decision tree is complete/complex enough, you will find that for any hand you play, you will only use 5-15% of the tree. This makes it unwieldy for constant use. I have found it best used to analyze a previous decision, after the fact. This is still helpful because it identifies possible flaw in my at-table logic.
The real problem with a single decision tree will be who decides if it is correct. I suspect that Daniel Negreanu's decision tree result might have a different answer than Phil Helmuth's. Though, they will probably be the same answer 80+% of the time. An individual strategy can't help but becoming reflected in the tree, otherwise we will have no context for which branch to go to next.
You mentioned "purchasing a guide" on this subject - what was it you purchased? Did it have a tree?

Good luck and God bless

#3 July 14th, 2020, 6:46 PM

Originally Posted by NWPatriot

You mentioned "purchasing a guide" on this subject - what was it you purchased? Did it have a tree?

Good luck and God bless

It was the PFE (post flop engine) from Upswing. It's very basic but effective. It doesn't cover everything obviously but gets you started with a categorization process that helps with each street. It's dynamic in that you re-quantify your hand with each street, up or down and act accordingly. This is of course (the acting accordingly) the unquantifiable bit in my opinion.

I heard there are more factors in hold em than there are atoms in the universe, which I suppose is true if you account for not only the probability of a given card, but the conditions in which you are playing (number of opponents, stack sizes, bet sizes, position, etc.). Not to mention, there's a psychological factor, which as far as I know isn't even close to being quantifiable even with a cluster of super computers!

I was more curious if anyone had a workflow for their decisions during a hand, with the understanding that this is mutable as the hand progresses. More of a curiosity in the end I guess. Although, now I am partially interested in perhaps developing some kind of neural network application to see if I can create my own program to determine these things. But who knows how long that will take! Decision Trees

#4 July 14th, 2020, 7:51 PM

Good decisions are the essence of poker.

The cards (ours, the board, our opponents) are important. Our reads are important. But in the end, these things are just information, and we must make a decision based on that information.

This knowledge doesn't make it any easier, but it does put it into perspective.

#5 July 14th, 2020, 11:10 PM

i could possibly make it a flow chart, but i ask myself general questions when making my decisions.

for example, preflop, i might ask myself:
-why would i limping in here?
-can i raise? will i call a 3bet?
-if i limp, will i call a preflop raise? can i 3bet?

if i'm betting
-does the texture of the board help my range or my opponent(s)?
-am i betting for value, a bluff, or to deny equity?
-what hands are calling with that i beat?
-what hands are calling with that beat me?
-what hands are folding that i beat?
-what hands are folding that beat me?
-what do i do if i'm 3bet?

if i'm facing a bet:
-what hands do i beat?
-what hands are my opponent betting that beat me?
-what hands that can beat me do i block?
-what hands can my opponent be bluffing with?
-what bluffs do i beat? what bluffs do i block?

it is a lot of questions, but i'm trying to ask myself those questions as the hand progresses to make the river decision a lot easier. basically, and usually at the river, if most of the answers are in my favor, i will bet/call a bet and if a lot of the answers arent in my favor, i'll fold. if it's pretty even, then it might be player depended, history with that player, or i just based on how i feel about the situation.

keep in mind that this isn't for every single hand, just the really big decision ones. i think, majority of the time, it's pretty clear cut bet/call/fold situations.

#6 July 15th, 2020, 1:19 PM

There's a lot of software for this, but it usually has a steep learning curve and can be expensive. Cardrunners EV would build complete decision trees and analyze hands this way. Piosolver also shows decision trees.

I think most top players, outside of software analysis, use the process more informally when they play just running through different possible outcomes along with their likelihoods.

#7 July 15th, 2020, 4:16 PM

I can simplify the game and create a decision tree, but it will be necessary to have enough poker knowledge to deal with unforeseen circumstances.

1- POSITION, GOOD RANGE, STACK: you need to have at lest two of this.

2- For each combination of the last itens you will need to take especifics choices, but for simplify use this:
a) POSITION + GOOD RANGE: play agressive.
b) POSITION + STACK: good for steal the blinds, you can see a flop and fold for a bet, here you can try some bluff.
c) GOOD RANGE + STACK: here you need to raise, and avoid to play against many players the flop, it no position you had to think twice every decision, becarefull with coordenated, suited flops or repeated cards on the board. Don't get attached to the cards.
3- PLAY ATENTION ON AGRESSOR PRE FLOP, avoid donk bets. if you are the agressor you can make a Cbet, but you need to read your adversaries and the flop.
4- THE FLOP, see the texture, look at the positions of the players and what are their range, don't raise every flop, try to show a good image at the table.
5- TURN, on this part the favorite hand will win in the majority of the cases, you need to reevaluate your hand, to spend money on marginal hands without good reads of the villain.
6- RIVER, here can happens miracles, you great hand can loose, sometimes just check will be ok.
7- END OF THE HAND, it's time to take notes and revise your gameplay.

This is basically the way I try to play, without going into too much detail or specific circumstances.

Similar Threads for: Decision Trees
Thread	Replies	Last Post	Forum
$2 NLHE 6-max: tptk turn decision	13	October 31st, 2019 6:26 AM	Cash Game Hand Analysis
JJ hands \| facing turn and river decision	5	October 12th, 2019 8:13 AM	Cash Games
€5 NLHE 6-max: KTs - right decision on flop?	8	October 1st, 2019 2:55 PM	Cash Game Hand Analysis
Decision Making	10	September 24th, 2019 11:54 PM	Learning Poker
Tough decision with pocket 10s on wet board	6	September 23rd, 2019 4:29 AM	Cash Games

Similar Threads for: Decision Trees > Texas Hold'em Poker
Thread	Replies	Last Post	Forum
$2 NLHE 6-max: tptk turn decision	13	October 31st, 2019 6:26 AM	Cash Game Hand Analysis
JJ hands \| facing turn and river decision	5	October 12th, 2019 8:13 AM	Cash Games
€5 NLHE 6-max: KTs - right decision on flop?	8	October 1st, 2019 2:55 PM	Cash Game Hand Analysis
Decision Making	10	September 24th, 2019 11:54 PM	Learning Poker
Tough decision with pocket 10s on wet board	6	September 23rd, 2019 4:29 AM	Cash Games

Decision Trees

Why more than 329,813 poker players have joined CardsChat