Memory in Poker

Comments off 697 Views0


Since the release of our preflop opening ranges, both in Level 1 of CORE and as part of our ongoing GTO expansion in PRO, I’ve noticed a growing trend. In CORE in particular, I now get frequent questions about whether subscribers should “memorize all of this” before moving on to Level 2. The follow-up question is often: “If so, how?”

Discussions with Vegas grinders have also revealed a certain level of gloom at the prospect of excellent memory being required for poker success in the solver era. With software tools revealing the objectively correct preflop and postflop play, will no limit hold’em be reduced to a memory contest between players?

There’s no doubt that when playing on poker websites, having good information recall is helpful. This is true of many strategy games that allow a limited time for decisions. And while I have no doubt that there are plenty of approaches to help you commit opening ranges and postflop lines to memory, probably involving something really depressing like flashcards, in this article I want to offer a different approach. Specifically, I’ll use some common poker situations to explore how much information you need to memorize and to what level of precision.

Opening Ranges

One of the first poker books I bought was “Hold’em Poker” by Sklansky and Malmuth. The analytic approach appealed to me greatly, as did the fact that playable opening hands were placed into eight distinct groups, with advice on which groups could be played profitably under different conditions.

Position is, of course, the most important variable, but the authors also explained how game texture impacts opening ranges. They even advocated what we would today call “mixed strategies”, with certain hands being played some fraction of the time.

It may be that I’m a sloppy student, but I don’t recall obsessing over my ability to memorize these ranges perfectly when first attempting to apply them to live play. I simply did my best. Further, the fact that opening ranges had a dependence on game texture and opponent tendencies revealed to me that, despite the science, good preflop play was also part art. 

This introduces an important question that I’ll lean on throughout this article. There is no doubt that, as poker players, we need to remember things like ranges. But how accurately must our recall conform to the charts offered on our site and elsewhere?

Here’s a trivial example that develops the point. There are a large number of preflop probabilities that players pick up fairly rapidly. We know, for example, that in a pair-versus-pair preflop match-up, the higher pair is a 80/20 favorite. Similarly, we are aware that two overcards are a coin-flip against a pocket pair.

Both of these pieces of conventional wisdom are approximations. In the case of the overcards-versus-pair scenario, the approximation is fairly poor; JJ is a 57/43 favorite over AKo, for example. If you tell a blackjack player that a 57/43 edge is roughly a coin-flip, they will conclude you’re an idiot.

The point is that we remember these common probabilities simply because they are good enough to arrive at sensible poker decisions. The fact that JJ vs 22 is a bigger favorite at 81.87% than JJ vs 66 at 80.96% simply doesn’t matter when we’re deciding how to play our hand. In fact, the reason we default to referring to these match-ups as 80/20s, rather than quoting probabilities to four significant figures, is a tacit acknowledgment that only a modest level of precision is required.

Does the same argument apply to opening ranges? With some qualification, I’d suggest it does. If you can memorize our charts to the point you know J9s is an open-raise from a certain position, but J8s is not, that’s great. But don’t kill yourself over it. It’s a classic case of diminishing returns. Yes, you might be able to develop perfect recall of a 6-max CO open-raising range by staring at the damn thing for another ten hours, but you’d likely generate far more EV using those ten hours to study a postflop facet of poker theory instead.

Besides, there’s a more important variable that adds slop to any opening range, to which I’ll now turn.

A Tournament Example

I started specializing in tournaments around fifteen years ago when LHE cash games started to dry up. Consequently, I spend a lot of time using tools like HoldemResources Calculator (HRC) to improve my own play and to explain tournament concepts to students.

Here’s an example of direct relevance to the current topic. Consider a 9-handed table in which everyone has 10bbs. There is a 0.1bb ante. HRC will happily find the Nash equilibrium for all players with a push/fold restriction. The UTG open-shove range is:

15.2%, 33+ A7s+ A5s ATo+ K9s+ KQo Q9s+ J9s+ T9s 

Like other GTO applications to poker, that range assumes our opponents are playing perfectly in a game-theoretical sense, which in this context means they are calling with unexploitable ranges. I won’t write them all out, but roughly UTG+1 should call with a 7.5% range. For the first caller, that number increases as one moves to later positions until it reaches about 10% for the player in the BB.

HRC also allows the user to explore scenarios in which calling ranges deviate from Nash. We can, for example, make all the single-call ranges 7.5%, then ask how that impacts the UTG shove range. The answer is UTG’s range increases from 15.2% to 24.3%. A fairly modest change to the calling ranges has thus produced a significant change to the shove range. If we instead stipulate that players call with the 5% range TT+,AJs+,AQo+, we discover that UTG should be shoving a whopping 86.4% of hands!

While some find explorations like this extremely troubling, I regard them as liberating. I remember shoving ranges fairly well simply through repetition, but the big takeaway from calculations like this is that game texture and the tendencies of specific opponents can completely swamp one’s baseline GTO ranges. Thus while one still needs to have a reasonable handle on the baseline ranges, profiling your opponents and understanding how they are responding to the changing tournament conditions is far more important than remembering if the MP2 shove range includes A7o+ or A9o+.

Memorization Has A Context

Another facet of this whole memorization issue is that we are not attempting to remember things in a complete vacuum. Particularly as we play more and study all aspects of poker, we can minimize the degree of memorization required simply from general poker principles.

In a preflop context, we don’t need to “remember” that 94o is outside all but extreme shoving ranges, because we know that 94o is not a good hold’em hand. Similarly, we don’t need to consciously include premiums in a shoving range.

Another simple example stems from our understanding that poker is positional. A shoving range from the button will be wider than one from UTG. A rudimentary knowledge of preflop equities also informs our decisions and assists memory. If we deduce that our opponents are shoving extremely tight, preflop equities demand that our calling range must be narrow. 

To Err Is Human

We may be living in the poker era of The Machines, but you’re not one. And stressing about playing like one strikes me as wasted energy. Yes, for restricted problems, we can now ask GTO+ or PioSolver to tell us the “correct” line for a specific hand on a flop of our choosing. And through cunning heuristics and simplifications, a dedicated student can implement strategies that are close to objectively ideal.

But recognize that all these tools, from opening charts to number-crunching solvers, are simply a means to better understand the beautiful game of poker. They will help you eliminate the big mistakes from your game first, then smaller ones, until you can dominate your player pool. But you’re never going to be perfect.