Statistical Arbitrage & Market Microstructure

Overview

Statistical arbitrage is a quantitative approach to trading that seeks to profit from temporary mispricings between related financial assets. Rather than relying on fundamental analysis or macroeconomic views, it draws on probability theory, time series econometrics, and systematic execution to identify situations where historical statistical relationships have broken down and are likely to revert.

At its core, the field rests on the concept of cointegration: the idea that certain pairs or groups of assets share a long-run equilibrium, even if their short-run prices diverge. When that divergence becomes statistically significant, an opportunity arises. Capturing it reliably net of transaction costs and with proper risk management is the central challenge this department addresses.

Market microstructure provides the foundation for understanding how prices are actually formed and how trades affect them. Order book dynamics, adverse selection, bid-ask spreads, and market impact are not peripheral concerns: they determine whether a strategy that works on paper can be executed profitably in practice.

What we work on

Members of this department build end-to-end research pipelines for statistical arbitrage strategies. This includes cointegration testing using Engle-Granger and Johansen frameworks, dynamic hedge ratio estimation via Kalman filtering, and spread modelling based on the Ornstein-Uhlenbeck process. Strategies are validated with rigorous walk-forward backtesting that accounts for transaction costs and realistic execution assumptions.

On the microstructure side, we study limit order book data, measure price impact across different market conditions, and develop execution algorithms designed to minimise market impact. Members are also encouraged to contribute to ongoing research questions around the decay of statistical relationships over time and the effect of regime changes on strategy performance.

All work is conducted with a high standard of statistical rigour. We place particular emphasis on out-of-sample validation and the honest reporting of results, including negative findings.