Do Prediction Markets Match Option Prices? Bitcoin Threshold Evidence from Binance and Polymarket

Battle Count: 50

Highly relevant for quantitative trading. The paper directly addresses cross-venue pricing inefficiencies between prediction markets and option exchanges, identifies a persistent (4-hour half-life) and economically meaningful pricing wedge (5.6-6.3 percentage points), and demonstrates a delta-hedged arbitrage proxy that generates positive net PnL after transaction costs. The structural findings (wedge largest at low implied probabilities and long maturities) provide actionable signals for strategy design. The Deribit extension showing an 11 pp gap further highlights exploitable cross-market mispricing. However, the small sample size and marginal statistical significance of the backtest (p=0.053) temper immediate practical deployment.

Moderate complexity. The core methodology relies on standard Black-Scholes pricing, numerical implied-volatility inversion (root-finding), and well-known econometric tools (t-tests, HAC, block-bootstrap, AR(1), ADF). The delta-hedged arbitrage backtest requires computing option Greeks (delta, vega) and constructing a three-leg hedge (call, spot, cash). The main implementation challenges are: (1) precise contract matching across heterogeneous platforms, (2) handling Polymarket trade-level data and converting No trades to Yes-probability scale, (3) time alignment across venues, and (4) computing friction-adjusted confidence intervals. No machine learning or deep learning components are involved.

All data sources are publicly available (Binance EOHSummary archive, Binance spot bars, Polymarket trade-level data). The methodology is fully specified with analytical derivations in the appendix (Propositions A.1-A.5). Contract mapping is explicit with exact strike, underlying, and maturity matching. However, no code repository is provided, and the exact Polymarket data extraction pipeline is not fully detailed. The narrow sample (3 contracts, 287 observations) limits statistical power for some sub-analyses.