Multi-factor models · Dynamic allocation · Risk premia analysisTraditional factor models — market beta, size, value, momentum, quality — are built exclusively from price and accounting data. Our factor construction incorporates data-derived signals: complexity metrics, linguistic divergence, information density, and cross-reference strength. These factors capture dimensions of return variation that are independent of traditional factors, providing genuine diversification in multi-factor analysis.
Factor premia are time-varying and regime-dependent. The value premium expands during recoveries, momentum weakens during regime transitions, and quality outperforms during flights to safety. Our aggregate signals provide real-time indicators of factor regime — allowing us to tilt factor weights dynamically rather than holding static allocations through changing market conditions.
Each factor premium compensates for bearing a specific risk. The value premium compensates for distress risk; the momentum premium compensates for crash risk; the carry premium compensates for event risk. We model the joint distribution of factor returns and weight analytical effort to maximise diversified signal quality — ensuring that no single factor dominates the research output during adverse regimes.