Why Does Complexity Make Factor Models Fail in Equilibrium?

We examine the impact of complexity on asset models and pricing errors in equilibrium when rational, risk-averse agents have imperfect knowledge of the data-generating process. Our model yields three implications as complexity rises: 1) pricing errors of factor modles increase, 2) high Sharpe ratio strategies are more likely to exploit estimation error components, and 3) there is an increasing number of these strategies that do not span each other.
The model rationalizes several asset-pricing puzzles, including the limited explanatory power of factor models, the weak relationship between beats and average returne, and the proliferation of seemingly unrelated anomalies. We provide empirical evidence documenting high complexity in predictive characteristics and covariances and a low correlation between sophisticated strategies.