This chapter examines statistical arbitrage strategies from their conceptual foundations, departing from the classical notion of arbitrage as a cornerstone of modern finance theory, and investigates the implications of these strategies for asset pricing dynamics. Grounded in the assumption of short-term mean reversion in prices and largely structured around market-neutral portfolios, this mechanism diverges from the traditional definition of riskless arbitrage by dispersing risk across a large number of simultaneous positions.

The chapter provides a comprehensive literature-based discussion of the principal strategy types, including pairs trading, cross-sectional momentum, fundamental-based value and contrarian investing, volatility arbitrage, and multi-factor model-driven arbitrage strategies. Four methodological frameworks are reviewed: the distance approach, cointegration-based methods targeting long-run equilibrium, stochastic control approaches that directly model spread dynamics, and machine learning models addressing high-dimensional data.

Beyond its role as a return-generating mechanism, the chapter analyzes the cumulative effect of statistical arbitrage on market efficiency from an asset pricing perspective. Anomalies exposed through academic publication undergo systematic decay as arbitrage capital is deployed, thereby accelerating price discovery. Nevertheless, this process does not imply that markets attain perfect efficiency. Drawing on the limits of arbitrage theory, the chapter examines how transaction costs, funding liquidity constraints, and model risk collectively impede the arbitrage mechanism.

The findings indicate that statistical arbitrage, as one of the fundamental forces sustaining market efficiency, progressively erodes its own return margins by eliminating the very mispricings it exploits. The chapter concludes by offering theoretically grounded implications for both researchers and practitioners regarding the structural gap between the theoretical return potential of these strategies and the real-world frictions that constrain their implementation.