Adaptive liquidity for prediction markets via online learning

Prediction markets rely on liquidity to turn trades into informative prices, but existing mechanisms fix liquidity ex ante, creating a static trade-off between price responsiveness and worst-case loss. In a new paper, Enrique Nueve, Bao Nguyen, Rafael Frongillo, and Bo Waggoner propose a fundamentally different approach: treat liquidity selection itself as an online learning problem. Their mechanism mixes a family of cost-function markets via learnable weights, yielding a single adaptive market that preserves no-arbitrage, bounded worst-case loss, expressiveness, and positive upside.

Key to the approach is a hybrid structural risk signal—a per-round objective that quantifies the trade-off between price impact and inventory risk. Standard online learning algorithms achieve switching-regret guarantees relative to the best sequence of liquidity regimes in hindsight. Simulations demonstrate the mechanism adaptively shifts liquidity across regimes in response to both order flow and inventory dynamics. The work establishes a principled framework connecting prediction market design with online learning, offering a path to more efficient and resilient markets.