Multi-Signal Ensemble: A Voting Mechanism for CTA Strategies
1. Introduction Single-indicator trading strategies are easy to construct but fragile in practice: each indicator has regimes where it fails, and a strategy that relies on one signal alone will suffer extended drawdowns during those regimes. The natural remedy is to combine multiple signals, hoping that their errors are at least partially independent. This is the ensemble approach, borrowed from statistical learning, applied to CTA signal construction. This article develops the mathematical framework for multi-signal combination, focusing on two canonical mechanisms—linear weighting and majority voting—and analyzes the conditions under which adding signals improves rather than degrades performance. ...