Cross-Market Political Arbitrage with Fractional Kelly Sizing

How It Works Related political markets on Polymarket (e.g., a candidate winning a primary AND a general election, or state-level races that are logically linked) must obey probability constraints. For example, P(candidate wins general) ≤ P(candidate wins primary). When these constraints are violated, the bot identifies the mispriced leg and trades toward consistency.

Entry / Exit Conditions Entry requires: (1) a detected no-arbitrage violation across related markets exceeding transaction costs, AND (2) the specific signal having a verified positive expected value on an out-of-sample backtest. This dual filter is explicitly described as a survival-first design philosophy. Exit is at resolution or when the violation closes.

Verified Results No live trading results are reported; the bot runs paper-trading by default. However, the backtesting requirement with out-of-sample validation is a methodologically rigorous filter that elevates confidence in the edge thesis.

Implementation Notes Fractional Kelly sizing (typically half-Kelly or quarter-Kelly) reduces variance relative to full Kelly while preserving most of the growth rate benefit. The "survival over profit" framing suggests conservative risk parameters. The small repo size (350 KB) indicates a focused, possibly early-stage codebase despite the sophisticated described methodology.

Risks Political markets are subject to resolution ambiguity and rule changes that can break assumed logical relationships. Backtest overfitting is a major risk given the limited history of prediction market data. Fractional Kelly still requires an accurate edge estimate, which is difficult to obtain in thin political markets.