The three-cowboy duel 2.0
About this app
Rebuild 2.0 (preview): 3D truel with exact survival probability per strategy, Monte Carlo with confidence funnel, selectable turn order and a map of when firing into the air is optimal.
An exact solution via a Markov chain
In a truel, three shooters fire in turn until one is left. While all three are alive there are only three states: “X is up”. From there the game flows into a two-player final duel. If X and Y alternate and X starts, X wins with P = p_X / (p_X + p_Y − p_X·p_Y). From these building blocks, the survival probabilities follow as absorption probabilities of a Markov chain, with no random estimate needed. With a random turn order, the final-duel formula becomes p_X / (p_X + p_Y).
What the app shows
Three shooters stand in a 3D scene: you, B and C. B and C always aim at the stronger opponent. You choose your shot while both are alive: into the air, at B or at C. For each choice the app shows the exact survival probability and an endgame table: who falls first, who starts the final duel and how much each path contributes. The turn order can be switched between weakest first, strongest first and random. A map shades the best strategy across your hit rate and B’s.
Simulation versus exact value
The Monte Carlo panel runs 1,000, 10,000 or 100,000 duels per strategy. A dashed line marks the exact value, and a band shows the 95 % range of an estimate from n duels. The app works out the n at which the two best strategies separate significantly (z = 1.96). Try the starting setups 10 · 60 · 90, textbook ⅓ · ⅔ · 1 and “strong you” 70 · 60 · 90, and watch on the map where firing into the air loses its edge.
Frequently asked questions
- How do you compute the winning chance in a duel with alternating shots?
- If X shoots first with hit rate p_X and Y follows with p_Y, X wins with P = p_X / (p_X + p_Y − p_X·p_Y). The formula comes from a geometric series: X wins on the first shot, or after a round in which both missed, and so on.
- Is firing into the air always the best truel strategy?
- No. It pays off when you are clearly weaker than both opponents and the turn order is fixed. Once you are a better shot than B, aiming at C yourself is better. With a random turn order, firing into the air is never better than the best aimed shot, because the next shooter is drawn anew anyway.
- How many simulation runs are needed for a reliable result?
- The random error of a Monte Carlo estimate shrinks with 1/√n. Four times the precision therefore needs sixteen times as many runs. When two strategies are close, a very large number of runs is required to tell them apart significantly.
- How does this version differ from the first truel app?
- This version computes the survival odds exactly with a Markov chain and compares them with the simulation and its confidence band. It adds a selectable turn order, an endgame table, a strategy map and a 3D scene for single duels.
Subject: Data & decisions
More from Data & decisions
- Anchoring effect
- Benford's law
- Berkson's paradox
- Cherry-picking: base period
- Cherry-picking: base period 2.0
- The three-cowboy duel (truel)
- The three means
- The three means 2.0
A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps