Parimutuel Betting on Permutations

We focus on a permutation betting market under parimutuel call auction model where traders bet on the final ranking of n candidates. We present a Proportional Betting mechanism for this market. Our mechanism allows the traders to bet on any subset of the n x n ‘candidate-rank’ pairs, and rewards them proportionally to the number of pairs that appear in the final outcome. We show that market organizer’s decision problem for this mechanism can be formulated as a convex program of polynomial size. More importantly, the formulation yields a set of n x n unique marginal prices that are sufficient to price the bets in this mechanism, and are computable in polynomial-time. The marginal prices reflect the traders’ beliefs about the marginal distributions over outcomes. We also propose techniques to compute the joint distribution over n! permutations from these marginal distributions. We show that using a maximum entropy criterion, we can obtain a concise parametric form (with only n x n parameters) for the joint distribution which is defined over an exponentially large state space. We then present an approximation algorithm for computing the parameters of this distribution. In fact, the algorithm addresses the generic problem of finding the maximum entropy distribution over permutations that has a given mean, and may be of independent interest.

💡 Research Summary

The paper addresses the challenge of designing a tractable prediction market for ranking outcomes, where the outcome space consists of all n! permutations of n candidates. Traditional combinatorial markets either allow betting on arbitrary subsets of outcomes—leading to exponential complexity—or restrict bets to very narrow languages, losing valuable information. The authors propose a novel “Proportional Betting” mechanism that lets traders bet on arbitrary subsets of candidate‑position pairs. A trader’s bet is represented by an n × n binary matrix Aₖ; the payoff equals the Frobenius inner product Aₖ·M_σ, where M_σ is the permutation matrix of the realized ranking σ. Thus each matching (candidate, position) pair yields a unit payoff, and a bet covering many pairs yields a payoff proportional to the number of matches.

The market organizer collects all bets and must decide which bets to accept in order to maximize worst‑case profit. Directly formulating this as a linear program would require a constraint for every permutation (exponential in n). The authors show that the separation problem for this LP reduces to finding a maximum‑weight matching in a complete bipartite graph whose edge weights are ∑ₖ Aₖ(i,j) xₖ (xₖ being the accepted quantity of bet k). Since maximum‑weight matching can be solved in polynomial time, the ellipsoid method together with this separation oracle yields a polynomial‑time algorithm for the organizer’s problem.

More importantly, the dual of the compact formulation yields a set of n² “marginal prices” p_{ij}, one for each candidate‑position pair. These prices satisfy the price‑consistency constraints of a parimutuel market and are unique when a small amount of “starting order” (a regularization term) is introduced. The marginal prices fully price any proportional bet, despite the exponential number of possible outcomes.

Having only marginal information, the authors turn to a maximum‑entropy principle to reconstruct a joint distribution over all permutations that is consistent with the observed marginals. They show that the maximum‑entropy distribution has an exponential‑family form: q(σ) ∝ exp(∑{i,j} θ{ij} M_σ(i,j)), where the n² parameters θ_{ij} are chosen so that the expected value of each indicator M_σ(i,j) under q equals the marginal price p_{ij}. This representation is compact (only n² parameters) yet defines a distribution over the full n! state space.

Computing the θ parameters exactly is a convex optimization problem with no closed‑form solution. The paper presents an approximation algorithm that iteratively updates θ using gradient‑based methods (essentially solving the dual of the maximum‑entropy problem). The algorithm runs in (pseudo‑)polynomial time and can achieve any prescribed accuracy ε.

Key contributions:

1. Introduction of a flexible proportional betting language for permutation markets.
2. Polynomial‑time formulation of the market‑organizer’s worst‑case profit maximization via a compact convex program.
3. Derivation of unique marginal prices that suffice to price all bets.
4. Construction of a maximum‑entropy joint distribution over permutations using only marginal prices, with a concise n²‑parameter exponential family.
5. An efficient approximation algorithm for fitting the maximum‑entropy model, which also solves the generic problem of finding the maximum‑entropy distribution over permutations with given expectations.

The work bridges combinatorial market design, convex optimization, and information‑theoretic inference, offering a practical framework that can be extended to dynamic settings (e.g., sequential call auctions) and potentially improve liquidity, truthfulness, and information aggregation in real‑world ranking markets such as elections, horse races, or sports tournaments.