Rock Paper Scissors as a Study in Strategy

A game with no best throw

Rock paper scissors looks trivial, which is exactly why it is useful. Rock beats scissors, scissors beats paper, and paper beats rock. Nothing is the best choice, because each option has another that beats it. Game theorists call this a cycle of dominance, and it is a clear small case of a game where strategy is all about what the other player might do.

Score a win as +1, a loss as -1 and a draw as 0.

The table gives your result for each pair of throws, with your choice down the side and your rival’s across the top. Whatever you choose, there is a rival choice that beats you, so none of the three throws is a dominant strategy. No pure choice is stable either: if the rival were certain to throw rock, you would throw paper, and then they would want scissors, and so on around the cycle.

Why one-third each is the equilibrium

The stable answer is a mixed one: throw each option with probability one-third. Check that it works. Suppose your rival mixes equally. Then for any throw you make:

• If you throw rock, you win against scissors (1/3), lose against paper (1/3) and draw against rock (1/3). Your average is (1/3)(+1) + (1/3)(-1) + (1/3)(0) = 0.
• The same sum applies to paper and scissors, by symmetry. Each gives an average of 0.

Every throw is equally good against an equal mix, so you cannot improve by changing your own plan, and neither can they. That is a Nash equilibrium, and by the symmetry of the game it is the only one. The average result for each player is zero, which makes sense, since the game is symmetric and the winner’s gain is the loser’s loss.

This also tells you something uncomfortable. Against a rival who plays the equal mix perfectly, nothing you do can give you a long-run advantage. You can neither win nor lose on average. The equilibrium mix is a guarantee of not being beaten, not a way to win.

Where the edge comes from

If an opponent is not perfectly random, the situation changes. Suppose a rival throws rock 40% of the time, paper 30% and scissors 30%. Your options are:

• Paper: wins against rock (0.4), loses to scissors (0.3), draws with paper. Average = 0.4 - 0.3 = +0.1.
• Rock: wins against scissors (0.3), loses to paper (0.3), draws with rock. Average = 0.
• Scissors: wins against paper (0.3), loses to rock (0.4), draws with scissors. Average = 0.3 - 0.4 = -0.1.

Throwing paper gains an average of 0.1 per round. The edge is small and builds slowly, since the biases are mild. It shows a general principle: you profit from a rival’s departure from equilibrium only to the degree they depart.

The risk of exploiting

Exploiting a pattern means leaving the safe mix, and that creates exposure. If you throw paper every time because you think the rival favours rock, a rival who notices can switch to scissors. A good approach is to move only as far as the evidence supports, and be ready to retreat. The more you commit to a read, the more you lose if it is wrong.

This is the same trade-off in many hidden-information games. Following the safe mix gives you a floor. Deviating toward a read gives you a higher possible gain with a bigger downside.

What people tend to do

Human players are not random, and some tendencies are commonly discussed, such as reluctance to throw the same option many times in a row, or a habit of changing after a loss. These are tendencies, not laws, and they vary with the person, the stakes and the setting. Treat them as hypotheses to test rather than facts.

It is easy to see patterns that are not there. A few rounds are a small sample, and streaks occur in random sequences. The risk is described by confirmation bias: once you think a rival favours rock, every rock throw confirms it and every miss is explained away. Believing that after three rocks in a row, paper must be “due” is a version of the gamblers fallacy. With a truly random rival, the next throw has the same chance as any other.

If you want to look for a pattern, track the rival’s throws over many rounds, and ask what you would expect from pure chance. A shortfall or surplus of one throw type over a handful of games is often just noise.

What it teaches about other games

• Simultaneous choice rewards unpredictability. Whenever a game boils down to guessing what the other side will do at the same moment, as in many combat resolution systems, similar logic applies.
• Counters create cycles. Many strategy and card games have matchups where A beats B, B beats C and C beats A. In these, the best choice depends on the field, and in a population the shares of each type tend to shift over time. This is a feature of the cyclic structure, not of any one game.
• Opponent reading is a skill built on evidence. Reading rivals is useful only to the extent that their behaviour really is predictable.

Variants

A well-known five-option variant adds two more throws. If each throw beats two others and loses to two others in a symmetric pattern, the equilibrium is again equal mixing, with a probability of one-fifth for each option. The logic is the same: symmetry means no throw is better than any other against an even mix.

If a variant makes the payoffs unequal, for example by giving a bigger prize for a win with one particular throw, the equilibrium shifts, and it can be found with the indifference method described in the mixed strategies article. As the matching example there shows, the bigger prize does not simply mean throwing that option more.

Series and tournaments

Games are often played in sets. Over a set, adaptation matters: your rival learns from your throws and you from theirs. In a best-of series, series length sets how much time there is to detect and react, and also how much time you have to be read. Since the equilibrium mix guarantees no long-run loss, the practical question for a series is how far to move away from it.

Rock paper scissors is a toy, but its structure is the foundation for much larger cases. When a game has no stable best choice, the right question changes from “what is best?” to “how do I avoid being predictable?”