How to Win Rock Paper Scissors

•Pate Bryant

You win rock paper scissors by being less predictable than your opponent and noticing when they are not. Against a truly random player nothing works: every strategy wins a third, loses a third and ties a third over time. People are not random, though, and two habits are well documented:

  1. After a win, people tend to throw the same thing again.
  2. After a loss, people tend to switch to the throw that beats the one that just beat them.

So: after you win, throw the move your winning throw just beat. After you lose, throw whatever beats the move that just beat you. Then test yourself against the predicting AI below, which looks for exactly these kinds of patterns in your own throws.

Why random play can't be beaten, and can't beat anyone

If you pick rock, paper and scissors with equal chance and no memory, the other player's choice does not matter. Whatever they throw, one of your three options beats it, one loses to it and one ties, each a third of the time. That is why the equal mix is the game's Nash equilibrium: nobody can do better against it.

The same fact cuts the other way. A random player also cannot exploit anyone. Against a player who throws rock every time, random play still wins only a third of rounds. To win more than a third you have to stop being random and bet on a pattern, which is exactly what opens you up to being read.

What a real study found about human habits

In 2014, Zhijian Wang, Bin Xu and Hai-Jun Zhou published Social cycling and conditional responses in the Rock-Paper-Scissors game in Scientific Reports. They recruited 360 students of Zhejiang University, split them into 60 groups of six, and had each group play 300 rounds, with players randomly paired inside their group each round and paid according to the points they earned.

  • Overall choices were close to one third each. Rock was 36%, paper 33% and scissors 32% on average, so looking only at totals, players seemed almost random.
  • Winners tended to stay. After a win, a player was considerably more likely to repeat that throw than to switch to either other one.
  • Losers tended to shift in one direction. In most of the payoff settings tested, a player who lost was more likely to switch to the throw that beats the winning throw than to repeat or switch the other way. Lose with rock to paper, and the next throw leans toward scissors.
  • These were tendencies, not rules. The authors note that the measured behaviour still differs considerably from a strict "win-stay, lose-shift" strategy. The edges are real but modest, and they show up over many rounds, not every round.

Those conditional responses also produced the cycling in the paper's title: across a group, the popular throw tended to rotate around the triangle over time instead of sitting still.

How to exploit those habits

Here is what the two habits suggest you throw next. They are tendencies, not rules, so treat them as a nudge rather than a guarantee.

What just happenedTheir likely next throwYour throw
You won with paper against rockScissors (beats your paper)Rock
You won with rock against scissorsPaperScissors
You won with scissors against paperRockPaper
You lost with rock to paperPaper againScissors
You lost with paper to scissorsScissors againRock
You lost with scissors to rockRock againPaper

Two short rules cover the table. After you win, throw what your winning throw just beat. After you lose, throw what beats the throw that just beat you. After a tie, the study gives no simple rule to bet on, so throw whatever you would least expect yourself to throw.

Other patterns are worth watching for, even without a study behind each one: a player who always opens with the same throw, a player who avoids throwing the same thing twice, or one who cycles rock, paper, scissors in order. Any of those makes them predictable. The AI below spots the last two within a few rounds; an opening habit only shows up across many separate games.

Your best defence is the same advice in reverse. If your opponent seems to be reading you, go back to random play. You cannot lose ground on average while you do, and you give them nothing to learn.

Play against the predicting AI

The AI picks its throw from your earlier rounds before you choose, so it never sees your current throw. Your rounds stay in this tab and are gone when you leave or start over.

Choose rock, paper or scissors to play the first round.

Rounds
0
You / AI / ties
0 / 0 / 0
AI win rate
0%
Predictability
0%

Predictability is the share of rounds where the AI named your exact throw before you made it. Pure chance is about 33%.

How the predicting AI works

Before each round, the AI looks only at your earlier rounds and makes a prediction from four simple models:

  • Frequency: the throw you have used most.
  • Last throw: what you threw the other times you had just thrown the same thing.
  • Last two throws: the same, using your last two throws as the context. This is a short Markov chain.
  • Last throw and result: what you did the other times you had the same throw and the same result, which is how it catches the win-stay and lose-shift habits.

Each model is scored on how often it would have guessed your earlier throws correctly, and the best-scoring model with a clear guess makes the call. The AI then throws whatever beats that prediction. If no model has a clear guess yet, such as on the first round, it throws at random.

Predictability is the share of rounds where the AI named your exact throw before you made it. AI win rate is the share of rounds the AI won. A player who cycles rock, paper, scissors gets read almost every round, while a truly random player keeps both numbers near 33%.

The game keeps your rounds in memory in this tab only. Nothing is sent anywhere or saved, and it is gone when you leave the page or press Start over.

FAQ

Is there a way to always win rock paper scissors?

No. Against an opponent who plays at random, no strategy wins more than a third of rounds on average. You can only gain an edge against an opponent who has patterns, and that edge is a tendency, not a guarantee.

What is the best first throw?

There is no strong answer. In the Zhejiang University study rock was slightly more common overall, at 36% against 33% for paper and 32% for scissors, which makes paper a marginally better guess. The edge is small, and an opponent who knows it can counter it.

Does the AI cheat by seeing my throw?

No. It commits to its throw from your earlier rounds before you choose, and the tests for this page check that its throw is the same whichever move you make in the current round.

Why does my predictability go up when I try to be random?

Real random sequences contain repeats and runs. If you avoid them because they do not feel random, that avoidance is itself a pattern, and the last-throw models can pick it up.

If you like games where reading the other player matters less than reading the position, try how to force a draw against impossible tic-tac-toe or this Connect 4 strategy guide.