The poker landscape has evolved significantly since the advent of poker solvers and various softwares that can simulate certain scenarios and build optimal strategies. Players have invested enormous amounts of time and energy into studying these solver outputs and different situations in attempts to improve their game and glean what they can from these computer programs. 

I would remind players that using solvers during gameplay is illegal, but outside of that, they can be a great as a teaching tool either before going into a session or reviewing one afterwards.

And while these programs have important value in the improvement process, I want to talk about a few of the things that solvers don’t do that well and the potential pitfalls of applying their outputs incorrectly.  

Understanding How Poker Solvers Work

A prerequisite on using a poker solver should be an introduction of how the software comes to its outputs. A solver program is typically given specific inputs from humans; information like stack depth, preflop ranges, positions, potential bet sizes and then told to play itself until a certain precision of the strategy is met. 

Solvers excel in the speed with which they can simulate these hands of self-play and converge on an optimal strategy for each player, often able to play millions of hands in minutes. Once the computation is complete, players are able to browse through the outputs and see which hands prefer certain actions on certain streets and study the game tree.  

Potential Pitfalls of Poker Solvers

Now that we know how the solver works, let’s talk about a few of the nuances that I have seen over the years working with them. Most of these concepts revolve around the idea of accuracy or congruency between the theoretical solver strategy and what we actually see in real poker tournament play versus humans. 

One aspect players frequently miss is the sensitivity of the strategies in response to small changes of the input. For example, let’s say a player runs a simulation with standard preflop ranges for a single raised pot between the CO and the BB. Simple enough right? However, let’s assume the player in the BB is a tighter player, only 3-bets premium hands and folds about half of the standard BB defend range (a common player archetype). If we run a simulation using a computer generated optimal BB preflop range, the output will be immediately irrelevant to how this spot will play versus this player.  

While players adept at working with solvers can remedy this simple example by tweaking preflop ranges, the same idea persists on every street in every decision. Because the solver is building its strategy by playing itself, there is a lot of what we call “range interaction” in the strategy. The output it ultimately converges on will be based on how it plays holding certain cards that interact with its opponent’s range. However, for the solver, the opponent is itself. 

A common theme postflop is human players not bluffing at a high enough frequency compared to the computers. If a solver has built a strategy based on calling down or re-bluffing certain bluffs that it has seen from the millions of self-play, but do not exist in real human play, then the output is going to be problematic to apply. 

The “Chaos Effect” of Real Life Players

The other, more interesting aspect of poker that solvers miss is the humanity of the players. Anyone who has played a significant amount of poker, especially in the live arena, has seen some strange things. Poker players go on tilt and play erratically, have favorite hands they always play regardless of the situation, or can be affected by outside forces on certain hands. I lump all of these factors into a concept called the “Chaos Effect”. This basically implies that there is going to be a degree of uncertainty in what is going on in live poker versus humans. Additionally, the lower stakes and/or lower caliber player you are playing with, the higher this Chaos Effect will likely be.  

Accounting for this Chaos Effect is something that observant human players can do intuitively, but solvers have an incredibly difficult time simulating. Even among elite players there are still patterns and favorite strategies that can be observed and even exploited. This is one of the more beautiful concepts in poker and one of the things that keeps me coming back to the game.  

Even with solvers playing at a superhuman level, there are still certain strengths in the human player and areas where we are still stronger. 

To try your hand at legal US online poker, register with BetMGM.