Rock Paper Scissors Probability

The Basic Odds: One Throw

Against a perfectly random opponent, every throw of Rock Paper Scissors has exactly three outcomes, each equally likely:

• Win: 1 in 3 (33.3%)
• Lose: 1 in 3 (33.3%)
• Tie: 1 in 3 (33.3%)

There are nine equally likely ordered pairs when both players choose each gesture independently with probability 1/3. The first player wins in three pairs, loses in three, and ties in three. No gesture is mathematically stronger in this model. The cycle only requires that every gesture beat one and lose to one; paper beating rock does not depend on physical realism.

If ties are replayed and only the next decided throw counts, symmetry gives each player a 50% chance to win that decision. This conditional result assumes neither player changes strategy because of a tie.

Match Probabilities: Best of Three and Beyond

Competitive RPS is almost never a single throw. Counting only decided throws (replaying ties), against a random opponent:

• Win a best-of-3: 50%. You need 2 wins before 2 losses
• Win 2 throws in a row: 1 in 4 (25%)
• Win 3 throws in a row: 1 in 8 (12.5%)
• Win 5 throws in a row: 1 in 32 (about 3%)
• Win 10 throws in a row: 1 in 1,024 (about 0.1%)

Any symmetric best-of format leaves equal independent players at 50/50. To model an unequal matchup, assume a stable probability for each decided throw. If one player independently wins each decided throw with probability 55%, the binomial model gives about 59.3% for best-of-5 and 67.9% for best-of-21. Those values change when throw outcomes are dependent or the edge changes during the match; the tournament format alone does not create skill.

Is Rock Paper Scissors Fair?

Under the stated symmetric payoff and independent equal-random assumptions, yes. Each player has the same options and expected payoff, and the 1/3 mixed-strategy Nash equilibrium cannot be exploited. The game theory guide distinguishes that expectation from a guaranteed finite-match result.

When two people replay ties and use the first decisive outcome, equal random play gives the same 50/50 win probability as a fair coin. Courts and auction houses have used RPS in documented decisions, but that history does not prove the participants randomized perfectly. See the sourced famous decisions.

Where the 1/3 Rule Breaks: Humans

Human-play studies show that the independent equal-random model can be an imperfect description. Wang et al. (2014) studied 360 university students in randomly paired six-person groups for 300 incentivized rounds and modeled population cycling through outcome-conditioned responses. Dyson et al. (2016) studied 31 undergraduates for 225 rounds against an equal-random computer and found more switching after losses and draws.

Neither study establishes a universal opening throw or a fixed advantage for Paper. Their findings motivate opponent- and setting-specific tests in RPS strategy and careful treatment of population, protocol, and uncertainty in RPS psychology.

Against a known equal-random opponent, no gesture has an edge. Against a person, bias is a hypothesis to measure rather than permission to assume a pattern.

Probability in the Variants

Adding actions changes tie probability only under stated assumptions. In Rock Paper Scissors Lizard Spock, each of five actions beats two and loses to two. If both players choose independently and uniformly, matching actions occur in 5 of 25 ordered pairs, so the tie probability is 1/5 (20%). For any n-action variant, 1/n is the tie probability only when the two choices are independent, uniform, and ties occur exactly when the actions match.

Test the Odds Yourself

Use repeated play to compare observed frequencies and transitions with a prediction made in advance. Separate practice, human competition, and AI results, and report the number of decided throws before interpreting a deviation. For defined WRPSA aggregates, see the statistics page.