This paper presents a preliminary analysis of a pairs trading strategy that utilizes the Ornstein-Uhlenbeck (OU) process to model the spread between stock prices. The research compares this sophisticated approach to a naive pairs trading strategy, which relies on a rolling window to compute mean and standard deviation parameters for generating trading signals. The study aims to evaluate the effectiveness of the OU process in capturing mean-reverting behavior and its practical implications for trading performance.

The findings indicate that while the OU model successfully identifies signals and trends in the spread dynamics, it underperforms the naive model when assessed on a risk-return basis. This underperformance is attributed to factors such as non-stationary pairs, which violate the OU process's assumptions, and challenges in parameter tuning that limit its real-world applicability. The paper contributes to quantitative finance by highlighting the trade-offs between complex statistical models and simpler alternatives in pairs trading, suggesting that further refinement or hybrid approaches may be necessary to leverage the OU process's strengths effectively.