Mixed strategies

How to randomize your decisions in poker

Last updated: 26 July 2026

To randomize a poker decision, take a target frequency from your strategy — say “bet 30%” — pick an outside source of chance you can read at the table, split its range in that proportion, and act on whatever it shows. The point is to move the coin flip outside your own head, where bias and patterns live, so you execute a mixed strategy honestly.

Why you can’t just decide in your head

A mixed strategy asks you to take the same holding and play it one way part of the time and another way the rest — for example, bet a given hand 50% of the time and check it the other 50%. The trouble is that people are famously bad random number generators. Told to “bet about half,” you drift: you bet when you feel strong, check when you feel weak, avoid repeating the last action, and settle into rhythms. Each of those is a pattern, and a pattern is exactly what an observant opponent needs. If your bets always mean strength, folding to them is free.

Randomizing fixes this by outsourcing the choice to something you don’t control. You stop deciding and start reading.

The three-step method

1. Get the frequency from your strategy

Your target comes from study — a solver output, a preflop chart, or a worked-out heuristic. It looks like “raise 60% / call 40%” or “bet 25% here.” That percentage is the whole input; everything else is execution.

2. Map the frequency onto a random source

Pick something unbiased and quick to read, and carve its range to match. A number from 0 to 99 is the easiest to reason about: for a 30% action, 0–29 is “do it” and 30–99 is “don’t.” A clock’s second hand works the same way if you treat the face as a pie chart; card suits give you fixed 25% / 50% buckets. (The methods, side by side, are their own guide.)

3. Act on what it shows — and mean it

Read the source before you let yourself have an opinion, then take the action it assigns. The discipline is the product: the one time you override the number because this spot “feels different” is the one time you leak. If the spot genuinely is different, that belongs in the frequency you chose in step one, not in a live override.

Worked example. You’re on the river with a hand your study says to turn into a bluff 30% of the time. You glance at your random source and it reads 17. That’s under 30, so you bet. Next orbit, a similar spot reads 61 — over 30, so you give up. Same hand class, two different actions, and no opponent can tell which you’ll pick.

Think in ranges, not single hands

What actually has to be random is your range, not any one hand. If your strategy check-raises a flop 25% of the time, the goal is that across all the hands you get there with, about a quarter become check-raises — a proper mix of value and bluffs — not that one specific combo is agonized over. Getting the overall frequency right is what makes the range balanced; per-hand precision beyond that buys you little.

Common mistakes

Randomizing in your head. The whole point is an outside source. If the coin flip happens between your ears, it isn’t random.
Reading the source after you’ve decided. Once you have an opinion, you’ll unconsciously bend the mapping to fit it. Read first.
Overriding “just this once.” Exceptions are a pattern too. Put your reads into the frequency, then follow the number.
Using a source that’s too coarse. Card suits can only give you 25% or 50% splits. If your target is 30% or 70%, you need something finer.

Key takeaways

Randomizing means outsourcing a choice to an unbiased signal you can’t skew.
Target frequency → map it onto a 0–99 range → act on the reading.
Read the source before you form an opinion, and don’t override it live.
Balance the range’s frequency, not individual hands.

Frequently asked

Can you just randomize in your head?

No — people are poor random number generators. Asked to bet about half the time, you drift: betting the hands that feel strong, avoiding a repeat of the last action, settling into rhythms an observant opponent can read. Randomizing works precisely because the signal comes from outside your control.

Should you randomize per hand or per range?

Per range. If your strategy check-raises a flop 25% of the time, what matters is that roughly a quarter of the range you arrive with becomes check-raises, with the right mix of value and bluffs inside it. Per-hand precision beyond that buys little.

When should you read the random source — before or after you decide?

Before. Once you already have an opinion, you will unconsciously bend the mapping to fit it. Read the source first, then take the action it assigns.