Time-Series Momentum Strategy Across Asset Classes
A systematic cross-asset trend-following framework using trailing-return signals, volatility scaling, and long-horizon tearsheet analytics.
This project presents the design and backtesting of a cross-asset time-series momentum strategy built on the empirical observation that trends in liquid futures and major asset classes can persist over medium horizons. The framework is structured to capture those persistent moves systematically rather than discretionarily, translating historical price behavior into scaled position signals with clear risk controls.
The methodology begins with 12-month trailing returns as the core signal used to determine directional exposure across equities, commodities, and fixed income. Rather than relying on nominal position sizes, the system applies ex-ante volatility scaling so that risk contribution is more comparable across assets with very different return distributions. This produces a cleaner measure of signal quality and prevents high-volatility contracts from mechanically dominating portfolio behavior.
Performance evaluation is comprehensive and regime-aware. The strategy is tested over more than two decades of daily data, with full tearsheet outputs covering Sharpe ratio, Calmar ratio, maximum drawdown, rolling attribution, and period-specific behavior during major macro episodes. That allows the analysis to distinguish between environments in which trend-following thrives, such as prolonged macro dislocations, and those in which choppy reversals compress returns.
The project is also designed to surface the economic logic behind results. Instead of stopping at headline performance, it decomposes return contributions across asset buckets and documents the specific catalysts associated with major gains and drawdowns. This gives the strategy a stronger research foundation and makes it easier to assess whether performance is being driven by a durable cross-asset effect or a small set of isolated market episodes.
The final framework demonstrates a disciplined approach to systematic trend capture, combining transparent signal engineering with professional-grade analytics. Future extensions include faster and slower lookback ensembles, dynamic volatility targeting, and macro-conditioned overlays that adapt exposure to changing monetary and inflation regimes.