Classifying the situations with a winning strategy in the card game Peer Pressure

The card game Peer Pressure, a perfect-information variant of War introduced by Boris Alexeev and posed in Peter Winkler's Mathematical Mind-Benders, splits a deck of distinct-valued cards between two players who see both hands; in each battle both simultaneously play a card, the higher card is permanently discarded, the lower card passes to the battle's winner, and a player who runs out of cards loses. Alexeev and Tsimerman's American Mathematical Monthly note proved golden-ratio sufficient conditions for a player to have a deterministic winning strategy and resolved Winkler's asymptotic question — under random dealing the probability that either player can force a win tends to zero — but closed with an explicit open problem: classify the situations, i.e. the pairs of hands, in which a given player has a winning strategy. Their sufficient conditions are not necessary (for example, the equal two-versus-two position {1,4} versus {2,3} is a forced win for Alice by the note's small-case facts but satisfies neither hypothesis), and the five-card position {1,2,4} versus {3,5} is a stalemate, so all three outcome classes are nonempty and intricate.

Background

War is a two-player card game in which each battle compares the players' cards and the winner takes them. Peer Pressure is a perfect-information variant of War introduced by Boris Alexeev and posed by Peter Winkler in the final chapter ("Unsolved and Just-Solved") of his puzzle book Mathematical Mind-Benders: a deck of $n$ cards carrying distinct values is split between two players with all cards face up, and both players know the contents of both hands at all times. In each battle, the players simultaneously play a card of their choice; the higher of the two cards is permanently discarded from the game, the lower card is passed to its player's opponent (the battle's winner), and a player who runs out of cards loses. Since exactly one card leaves play per battle, a game lasts at most $n-1$ battles; since card values are only ever compared, only their relative order matters.

A player has a winning strategy if she may announce her entire strategy in advance and still always win against any strategy of her opponent. Strategies may in general be mixed (probabilistic), but if a mixed winning strategy exists then a pure (deterministic) one does too, so "winning strategy" is read as deterministic throughout. Because the two players move simultaneously, the game is not determined in the classical sense: a position can have no winning strategy for either player. With four or fewer total cards one of the players always has a winning strategy — the player with more cards if the hands are unequal, otherwise the holder of the highest card — but with five cards there is essentially one position, hands $\{1,2,4\}$ versus $\{3,5\}$ , in which neither player has a winning strategy.

Alexeev and Tsimerman analyzed this game in their American Mathematical Monthly note. They proved a monotonicity principle — replacing a player's cards by better ones, compared card-by-card from the top, never hurts her — and the following sufficient conditions. Writing $a$ and $b$ for the numbers of cards held by Alice and Bob and $\phi=(1+\sqrt{5})/2\approx 1.618$ for the golden ratio, Alice has a winning strategy if $a>\phi b$ , or if $b<\phi a$ and all of her cards are higher than all of Bob's. From this Main Lemma they resolved the asymptotic question Winkler had posed: under random unbiased dealing, the probability that either player has a winning strategy tends to $0$ as $n\to\infty$ (a limiting-probability-one statement, which the paper calls generic), and if instead one player receives $r$ cards for every card of the other, the limiting probability jumps from $0$ to $1$ at the threshold $r=\phi$ . These conditions are only sufficient: the small-case facts already yield winning positions satisfying neither hypothesis, for instance the equal two-versus-two position $\{1,4\}$ versus $\{2,3\}$ , which Alice wins as holder of the highest card although $a>\phi b$ fails and her lowest card is below Bob's. The note closes by posing the exact finite classification that remains unresolved.

Problem Statement

Classify the situations in which a given player has a winning strategy in Peer Pressure. A situation is a pair of face-up hands $(A,B)$ of cards with distinct values (only the relative order of the values matters, so the values may be taken to be $1,\dots,n$ ); the players simultaneously play a card each battle, the higher card played is permanently discarded, the lower card is passed to its player's opponent, and a player who runs out of cards loses; a player has a winning strategy if she may announce her entire strategy in advance and still always win against any strategy of her opponent (equivalently, by the paper's reduction, a deterministic winning strategy exists). Give an exact characterization — necessary and sufficient conditions on $A$ and $B$ — determining, for every situation, whether Alice has a winning strategy, Bob has a winning strategy, or neither player does. The Main Lemma's sufficient conditions ( $a>\phi b$ , or $b<\phi a$ together with all of Alice's cards higher than all of Bob's, where $\phi=(1+\sqrt{5})/2$ ) are not necessary; with at most four total cards some player always has a winning strategy (unequal hands mean the larger hand wins; equal hands mean the highest card's holder wins), while the five-card position $\{1,2,4\}$ versus $\{3,5\}$ belongs to neither player.

References

Boris Alexeev and Jacob Tsimerman, "Note on a War-like Card Game", American Mathematical Monthly 119(9), pp. 793–795, 2012. DOI: 10.4169/amer.math.monthly.119.09.793, https://doi.org/10.4169/amer.math.monthly.119.09.793 (the source of the open problem; the published text is paywalled).
Boris Alexeev and Jacob Tsimerman, "An analysis of a war-like card game", preprint, 2010. arXiv:1001.1017, https://arxiv.org/abs/1001.1017 (authors' preprint of the Monthly note).
Peter Winkler, "Mathematical Mind-Benders", A K Peters, Wellesley, MA, 2007. ISBN 978-1-56881-336-3, https://zbmath.org/5208110 (poses the Peer Pressure problem in ch. 11 "Unsolved and Just-Solved", p. 134).
Manan Bhatia, Byron Chin, Nitya Mani, and Elchanan Mossel, "When Will (Game) Wars End?", American Mathematical Monthly 133(1), pp. 35–46, 2026 (online November 2025). DOI: 10.1080/00029890.2025.2572283, https://doi.org/10.1080/00029890.2025.2572283; preprint arXiv:2302.03535, https://arxiv.org/abs/2302.03535 (cites the Monthly note; studies the termination time of War variants).
Tanya Khovanova and Atharva Pathak, "Combinatorial Aspects of the Card Game War", preprint, 2022. arXiv:2202.00473, https://arxiv.org/abs/2202.00473 (cites the note's preprint; combinatorial analysis of standard War).