This project demonstrates a critical insight: even a straightforward, intuitive RSI-based trading strategy can produce results that range from seemingly exceptional to utterly ineffective, based solely on one often overlooked variable, the day of the week used for backtesting.
This finding underscores a broader lesson for systematic investing and quantitative research: without rigorous robustness testing, our conclusions may be built on fragile, arbitrary foundations. In today’s data-driven landscape, reliability must come from stress-tested frameworks, not convenient assumptions.
The foundation of this analysis is a deliberately simple, rules-based strategy. Its purpose is not to be the "perfect" model, but rather to serve as a clean, controllable environment for testing the impact of signal timing. By keeping the logic straightforward, we can isolate and measure how a single variable shift can dramatically alter outcomes.
The simplicity allows us to isolate the impact of signal timing and stress-test how sensitive performance is to seemingly minor variables like day-of-week execution.
The analysis is structured around two phases: an In-Sample period for parameter optimization, and an Out-of-Sample period for performance validation.