Arbitrage Strategies under Weak Market Dependencies

Investigate arbitrage strategies in order-book prediction markets such as Polymarket that exploit weaker, temporal logical dependencies between distinct markets—for example, cases where the resolution of a preliminary event ("Team A wins the semifinal") constrains the outcome space of a subsequent market ("Team A wins the final"); determine the conditions and timing windows under which such strategies can be executed profitably.

Background

The paper defines two principal arbitrage types in prediction markets: Market Rebalancing Arbitrage (within a single market) and Combinatorial Arbitrage (across strictly dependent markets). Dependence is operationalized via subsets of conditions whose truth assignments are logically constrained across markets, enabling guaranteed cross-market arbitrage when summed prices diverge.

In practice, many market pairs exhibit weaker logical connections that do not meet the strict subset-dependence criterion but still create temporal windows for arbitrage—such as tournaments where the outcome of a semifinal affects the feasible outcomes of a final. The authors note that LLM-generated outcome tables reveal such weaker dependencies and explicitly mark the paper of strategies to exploit them as an open problem.