GTO Poker From Scratch. A Poker Solver Tutorial for First-Timers


If you're following our blog, you probably already know the benefits of studying the GTO poker strategy. If that's not true, we strongly suggest you start with the basics of game theory optimal and read what a poker solver is.

In another piece, we stressed how important it is to focus on a broader spectrum instead of getting overly attached to the results of a single solution. However, nobody is born a great poker player, and if you're new to studying GTO, you have to start your poker training somewhere.

In this article, we'd like to guide you through the thought process of analyzing hands with Deepsolver.

Studying game theory takes time, but it's essential to become an above-average poker player

If you've never used any poker solver, it may be quite a challenge to find your way through the maze of available options, especially given how rough the interface of many poker tools can be. That's why we'd like to take a real-life example and show you how you can break down a hand step-by-step, or as we should rather say, "street-by-street", using Deepsolver.

To explore the optimal strategy for any heads-up spot you like, start by choosing the "Custom Solution" option. From there, you have two choices - a custom or smart view. We'll go with the latter as it's more visual, but it's up to you which one you prefer.

Smart view of the custom solution
Smart view of the custom solution

On the left, there's a view of poker table and a few options: stack depth in big blinds, game format and positions of both players in the hand. For the sake of this example, let's assume that you'd like to improve at a troublesome spot for many novice poker players - how to approach the river in 3-bet pots after betting the flop and the turn out of position. We'll pick cash games at 100BB as our format of choice.

Let's investigate the strategy for the big blind playing against a player on the button. After choosing the desired positions and the 3BPOT option, Deepsolver will propose the game theory optimal ranges for both players. Keep in mind that these reflect how the preflop play should look like. It's almost certain that in poker games on lower stakes, both your and your opponent's range will differ heavily from what's supposed to happen.

However, by using the edit option, you can input any range you want. At the time being, we'll reduce the hands both players have to more manageable numbers.

Let's use tight ranges for both players; 10% of hands for BB and 16% for BU.

We'll be investigating this situation with those two exact ranges
We'll be investigating this situation with these two ranges

Choosing the available bet sizes and the board texture are the next steps

One of the biggest strengths of Deepsolver is that it will provide you with the GTO strategy for any heads-up play situation in a matter of seconds. While conducting a custom solution, you can choose a random board texture or pick a specific one. In this instance, let's pick a dry texture (meaning there are few draws on the board) of the J92 rainbow.

The last thing to set up is a decision tree. This tree consists of all the bet sizes that both players are allowed to use at any given point of the hand.

You have a lot of freedom while choosing the sizings used in solutions (up to five different ones on every street!). A proper approach to bet sizing is an extensive topic on its own, probably for another article or even a series.

To keep this example simple, we'll use two bet sizings for a first bet (one smaller - 33% of a pot and one bigger - 75% of the pot), one raise size (75% of a pot) and one re-raise size (also 75%). We'll leave the all-in options unchecked for the time being.

After some time spent with Deepsolver, you'll likely find your most preferable decision tree. You can also use a Smart tree option, but keep in mind it's still in beta.

The more bet sizes you input, the more advanced poker strategy you'll receive. Be aware, though, that when fed too much data, poker solvers can come up with very hard-to-replicate lines.

Also, remember that it's way better to execute a straightforward poker strategy very well than to use a complex plan incompetently.

We're good to go and calculate!
We're good to go and calculate!

Let's see what the optimal poker strategy looks like on the flop!

After several seconds, we'll receive our first results. And there's quite a lot of data to look at.

An optimal strategy for the out-of-position player

Our work starts with analyzing the numbers above the poker table visualization - labelled EQ and EV.

Equity (EQ) reflects the average number of chips you'd win if there were no more betting and the hand would reach showdown on the river (considering all possible outcomes for turn and river cards).

Expected value (EV) determines the long-term value of your range (or a particular hand). By following the solver guidelines, you'd win, on average, the number of chips presented by the expected value.

Why are these two numbers so important? Because they help you to get the overall grasp of what's going on in the particular moment of the hand - which player has a stronger range and by what margin. The inequalities in players' ranges dictate distinct strategies for different players. Usually, if one player has a clear equity advantage, they should be playing quite aggressively.

Let's look at our example: a player on the big blind has a strong preflop range, and almost all of their holdings on the chosen board have a decent draw, two overcards or a pair or better. In this scenario, only 76s and a few suited wheel combinations missed. On the other hand, when we look at the BU range, we notice that they have quite a few strong holdings like pairs, sets and decent draws themselves but also have even more weak hands like 76s, 65s, 54s and quite a few suited aces A8-A3s which are unlikely to improve significantly on further streets.

You can investigate the exact equity of all the possible holdings for both players by choosing the EQ option in the "Matrix" drop-down.

This discrepancy results in the out-of-position player having an equity advantage (in this case, 53% to 47%). At the same time, the big blind has almost all of the nutted hands in their range (strong pairs, all of the overpairs and sets of jacks and nines). The player on the button has all of the possible sets and best top pairs but lacks all the possible overpairs.

As a result, the out-of-position player can bet frequently and quite big (roughly 24% of their range using 1/3 sizing and a little over 55% with 3/4 sizing). Even if you've never used any poker software, you would likely bet this flop quite often after 3-betting preflop. In this case, this intuitive approach is correct.

The out-of-position player has chosen to cbet and get called. What's next?

Let's pick the big bet sizing (75% of the pot) and choose a call for the in-position player; how should we proceed? The next step is to select a turn card and to choose the bet sizings once again. Let's pick a queen of diamonds and leave the decision tree as it was (bets of 33% and 75%).

How does optimal strategy look now?

Out of position player's range still looks strong!
Out of position player's range still looks strong!

A quick look at the EQ and EV numbers should lead you to believe that big blind players should keep betting, and the colours on the hand matrix confirm this assumption. A queen is a good enough card for the out-of-position player to continue the aggression - it improves some of their hands (AQ, KQ) to top pairs, completes a few draws (KT, QT, T8) and doesn't weaken the flop overpairs.

The main difference is that the solver likes to bet 33% instead of 75% of a pot. Why is that? The primary reason is that even when betting 33%, the pot on the river will be big enough (over 80 chips with about 70 behind for each player) to get all the money in smoothly.

Of course, the above approach won't always be the case - the strategy for the out-of-position player will be different for other turn runouts. Can you guess what would the GTO strategy look like for a turned ten of hearts or deuce pairing the board?

We invite you to find out yourself, as you might be surprised by the results. Even if you're not a Deepsolver user yet, you can try it out with a seven days free trial!

According to game theory, should the big blind put more money on the river?

River is a unique point of every poker hand for a few reasons. Let's look at the two of them.

The first: the pot is the biggest on the river, so every mistake you or your opponent make will be much more costly than on earlier streets.
The second one: there are no more cards to come, so the board texture won't change - barring chops, your hand is either the best or the worst.

Due to the second factor, you can simulate and grasp many river outcomes relatively quickly, as you do not have to consider the implications of more cards coming on future streets.

Let's look at what happens when the river bricks with a three.

OOP still likes to bet a lot
Big blind should still bet often

As you can see, the BB player's range stays strong and maintains an equity advantage, resulting in a continuously aggressive strategy. As a result, the majority of the top pairs and better are clear valuebets. What's important: with so many hands that can bet for value, big blind can bet all of their AK combinations (which is probably the most troublesome hand for many novice players in this spot).

Let's look at the different outcome. What if the river was a jack of diamonds?

That's pretty bad outcome for a big blind player.
This card is a disaster for the big blind player

That's probably the worst card you could see being the out-of-position player. It drastically changes both the EQ and EV in favour of the player on the button, forcing the BB player to play very defensively and not bet much! The reason is simple, BU has way more jacks in their range than BB.

These are only two possible scenarios, but in a matter of minutes, you can try out all the possible river cards and prepare for almost any situation. Of course, you won't be able to memorize all the outcomes at once, but with a solid repetition, you'll find the patterns which will emerge, as the strategy for many cards will be similar.

That's only the introduction to the GTO poker strategy

But that's more than enough for a start. Bear in mind that you can tweak almost every variable mentioned during the hand analysis process. From the preflop ranges through stack sizes to bet sizes. That gives you nearly endless possibilities.

For now, if you are new to the poker solvers world, we suggest you stick to a limited number of options. With time, you'll easily incorporate more complex solutions into your play.

We also recommend taking a look at our What's the Best Poker Software To Develop Your Game article - where you'll find out about great free poker tool and What is an Optimal Poker Strategy? Game Theory Optimal Poker vs Exploitative Approach which will shed some light on the GTO poker vs exploitative strategy debate.

A few years ago, solvers revolutionized how we learn, study and play poker, and we believe that Deepsolver is a game changer on the market. It doesn't need a lot of computing power, has an intuitive interface and is easy to use.

But don't take our word for it; check Deepsolver out yourself. If you have any questions, ideas or want to connect with our other users, our Discord is the place to go.