Mixed Strategies: One Hand, Two Ways

Open a solved preflop chart and some cells refuse to give a straight answer: the same hand is a raise part of the time and a call the rest, or a raise blended with a fold. To a player raised on rules of thumb this looks like indecision, or worse, like the solver hedging. It is neither. Mixed strategies are one of the most informative things a solve produces, and understanding why they exist changes how you read every chart you own. This piece covers where mixes come from, why they cluster at the edges of ranges, how humans can actually execute them, and when rounding them off is cheap versus costly.

Indifference: the engine behind every mix

A solver never mixes out of uncertainty. It mixes because, at equilibrium, the hand in question earns exactly the same expected value from two different actions. That condition is called indifference, and it is not an accident of the math, it is a product of it.

Here is the intuition. Suppose a hand strictly preferred raising to folding in some spot: raising earns more. Then the solver would raise it always, and the hand would be pure. The only hands that end up split between actions are the ones where the alternatives have been driven to exactly equal value. And what drives them equal is the opponent's strategy. At equilibrium, your opponent defends at frequencies that make your marginal hands indifferent between their options, and you do the same to theirs. Neither side can be pushed around, because the hands that could be pushed have nothing to lose either way.

That is worth sitting with, because it inverts the amateur reading of a chart. A mixed cell does not mean the solver could not make up its mind. It means the two players' strategies have squeezed that hand to the exact point where the decision, for that hand alone, no longer matters. What still matters, enormously, is the frequency.

Why the frequency matters when the EV does not

If a hand earns the same whether it raises or calls, why does the chart bother specifying how often to do each? Because the frequency is not for the hand. It is for the range.

Your opponent does not play against your hand; they play against the distribution of all hands you take a given action with. Every boundary hand you move from calling to raising changes what your raising range looks like: a little wider, a little weaker, a little differently constructed. Individually the moves are free. Collectively they reshape the range, and a reshaped range invites a counter-strategy. If every hand that was supposed to mix between raising and folding simply folds, your raising range gets stronger but rarer, and opponents can start folding correctly against it and stealing the pots you no longer contest. If they all raise, your raises get frequent but weak, and playing back at you gets profitable.

The mixed frequencies are the calibration that keeps every counter-strategy neutral. They make your raising range exactly strong enough, your calling range exactly wide enough, and your folds exactly rare enough that no adjustment against you gains anything. The individual hand is indifferent; the range is anything but.

Pure strategies own the core

None of this means solved poker is a haze of coin flips. The overwhelming majority of decisions in any solved range are pure. Premium hands raise every time, because raising strictly beats the alternatives and indifference never enters into it. Hopeless hands fold every time for the same reason in reverse. One action dominates, and the solver takes it without ceremony.

Mixes live where domination runs out: at the boundary between playing and folding, or between two ways of playing. Picture any solved range as a territory. The interior is settled, pure, and frankly boring. The frontier is where the interesting cells sit, hands good enough that folding them always would waste value, but weak enough that playing them always would overextend the range. The chart's blended cells trace that frontier precisely. If you have not already, read how to read a preflop chart, because the split cells that confuse new chart readers are exactly this frontier made visible, and once you see them as the boundary of the range rather than noise inside it, the whole grid becomes legible.

This geography also tells you where your study time goes. The core takes care of itself; nobody needs a chart to open aces. The edges are where solved output earns its keep, because the edges are precisely where intuition has nothing to grab onto.

What a mix protects

The defensive job of mixing is easiest to see in 3-betting. A 3-bet range needs weak hands in it, not because weak hands are secretly strong, but because a 3-betting range made only of monsters is a range nobody pays off. Opponents fold everything but their own monsters against it, and your premiums, the whole point of the range, stop getting action. The mixed 3-bets at the bottom of the range are what force opponents to keep calling and 4-bet bluffing, which is what keeps your value hands profitable.

The same logic runs through every action. Mixing calls into a spot where you also raise keeps your calling range from being capped at mediocrity, so opponents cannot safely barrel every board against it. Mixing between opening and folding at the bottom of an opening range keeps your steals frequent enough to deny the blinds an easy life without making your average open weak. In every case the pattern is identical: the mix spends a near-zero-EV hand to buy protection for the high-EV part of the range. Zero EV hands are the cheapest insurance in poker, which is why the solve buys so much of it.

The human problem

All of which is elegant, and none of which addresses the awkward fact that you are not a solver. A solver executes a mixed frequency exactly. A human asked to do something a fraction of the time, with no aid, does no such thing: we streak, we compensate, we drift toward whatever felt good last time it worked. Worse, our deviations are not random, they are correlated with mood and momentum, which makes them readable.

The fix is to outsource the randomness to something external. Live players have a device already in their hands: suits. The suit combination of your hole cards divides naturally into quarters, so a rule like taking the aggressive option only with one named suit gives you a workable approximation of a quarter frequency, and variations on the idea cover other splits. Online players have easier options: the second hand of a clock, the last digit of a timer, or a dedicated randomizer app, consulted before acting. The specific tool is unimportant. What matters is that the decision is made by something that cannot tilt, cannot remember the last three orbits, and cannot be read by the player across the table.

There is also a legitimate lazier path, which is not to randomize at all, and it deserves an honest treatment rather than a dismissal.

When simplifying costs little

Because mixed hands are indifferent, rounding a mix to a pure strategy is cheap in the spot where you do it. Take the action the solve favors most often and play it every time: in that node, against that strategy, you have given up almost nothing, since the actions were near-equal in value to begin with.

Simplification costs least when the conditions that made the mix matter are absent. In rare spots, nobody accumulates enough of a sample on you for the imbalance to be found, let alone attacked. Against anonymous or inattentive opponents, an exploitable pattern that no one is tracking is not exploited. And for a hand deep in mixing territory, where the frequencies are lopsided anyway, promoting the heavy side to always is barely a deviation at all. For most players in most games, a chart simplified to majority actions captures nearly all of the value of the full solution, and is dramatically easier to execute under pressure.

When it costs a lot

The bill arrives when simplifications pile up in the same direction. Each rounded mix is free; a systematic lean is not. Resolve every borderline open toward folding and you have not made one cheap simplification, you have made your whole late-position strategy tighter than the solve, and the blinds can respond by folding more to your now-stronger opens and re-stealing wherever you fold. Resolve every marginal 3-bet toward calling and your 3-bet range drifts toward value-only, which is the exploitable shape the mix existed to prevent. The individual roundings were near-zero EV; the aggregate lean is a leak with a shape opponents recognize.

Frequency of the spot multiplies the cost. A simplification in a node you play a handful of times a year will never be discovered. The same simplification in a constantly recurring battle, against regulars who see you every day, is a standing invitation. The rule that falls out: simplify by rounding each mix to its own majority action, never by resolving whole classes of mixes toward the passive or the aggressive option because it feels safer, and stay closest to the true frequencies in your highest-traffic spots against your most observant opponents.

Owning the exact frequencies

Every path through this topic ends at the same requirement. Whether you intend to execute the mixes faithfully with a randomizer, or simplify them intelligently and need to know which side of each mix is the heavy one, you need the actual frequencies, not a memory of a simplified chart someone posted years ago. The poker preflop charts show every mixed cell as a live split with its real frequencies, spot by spot, so you can see exactly where the pure core ends and the frontier begins. And the Master Bundle puts the full solved ranges in your hands permanently, open in the Simple Preflop solver, for every format and depth. The solve already did the impossible part, driving thousands of boundary decisions to perfect indifference. Your part is only to know where those boundaries are and decide, deliberately, how faithfully to walk them.