Structural Modeling and Forecasting of Volatility in Binary Prediction Market Contracts

01 ABSTRACT

Prediction market prices are bounded probabilities with binary payoffs and known deadlines, making standard volatility models a poor fit. The authors develop a structural volatility model with a Wright-Fisher deadline-resolution component that captures forced binary uncertainty resolution over time and a Glosten-Milgrom order-flow component that captures informed trading via spreads and volume. Using a large panel of Kalshi contracts, structural variables show substantial forecasting power, dominating plain ARCH/GARCH benchmarks, with the best overall forecasts from scaling structural variance by a GARCH multiplier.

02 KEY FINDINGS

1. The model decomposes prediction market volatility into a deadline-resolution component and an order-flow component, capturing time-driven and trading-driven uncertainty respectively.
2. On Kalshi contract panels, structural variables significantly outperform plain ARCH/GARCH benchmarks, with the structural term setting the conditional variance level and the GARCH recursion correcting the timing of interval violations.
3. Volatility is highest near fifty-fifty prices, rises near resolution, and varies across categories with differences in the timing and discreteness of information arrival.