Verification of profitability of statistical arbitrage on the US stock market

This thesis primarily investigates the pairs trading strategy and its ability to generate excess returns over the broad market, measured by the S&P500 Index, during variable market conditions. The analysis is done on a large portion of the US stock market, using the stocks included in the Russell 3000 Index as a data source. The strategy is back-tested over four distinct periods from 2020 to 2023, each characterized by varying market conditions, such as economic recession and high volatility periods, but also stable markets. The pairs are identified using the two-step correlation and cointegration method. The transaction costs, like transaction fees, bid/ask spread, and slippage, were accounted for to present more accurate outcomes. The evaluation is based on several predefined evaluation metrics, such as annualized excess return and annualized risk-adjusted returns measured by the Sharpe and Sortino ratio, as well as the number of transactions and their success rate. These metrics are put in place to make it easier to compare the different variations of the strategy. After the baseline evaluation, the strategy is slightly altered, with several risk mitigation techniques or changes in the input parameters implemented to evaluate the individual impacts of each factor on the strategy’s performance and test its robustness and stability. Lastly, two hypotheses are formulated and empirically tested in an attempt to explain performance.