Investment Thesis

In quantitative finance, there has long been a strong emphasis on parameter optimization to improve strategy performance. However, extensive testing on out-of-sample data reveals that over-optimization often leads to poor generalization and suboptimal real-world results. Interestingly, our analysis shows that the most robust performance comes not from a single optimized parameter set but from an averaged approach across multiple parameter variants.

This algorithm employs a momentum-based trading strategy that averages position sizing across four distinct parameter configurations. Position sizes are recalculated dynamically, with signal strength determined by the distance between moving averages—providing a nuanced measure of market momentum.

To enhance risk management and robustness, the strategy incorporates several protective mechanisms:

  • Stops: To limit downside risk on individual trades.
     
  • Market Regime Filter: In bearish market conditions, the portfolio automatically reallocates to bonds, preserving capital during downturns.

 

By balancing diversification across parameters and integrating safety measures, this approach seeks to achieve consistent performance and adapt to varying market conditions while avoiding the pitfalls of over-optimization.

Author

Quant League Competitions

January 2025