Consensus-Bottleneck Model Improves Stock Return Prediction
The Consensus-Bottleneck Asset Pricing Model (CB-APM) introduces a novel approach to asset pricing by incorporating aggregate analyst consensus as a key structural bottleneck, considering professional opinions as a sufficient statistic for market data. This model is designed for interpretability, utilizing the bottleneck constraint as an endogenous regularizer, which enhances out-of-sample predictive performance and aligns inference with economically meaningful factors. Portfolios based on CB-APM predictions exhibit a strong, consistent return gradient across various macroeconomic conditions. Additionally, pricing diagnostics indicate that the consensus captured reflects priced variations overlooked by traditional factor models, highlighting belief-driven risk differences that standard linear frameworks fail to account for. The full paper can be accessed on arXiv.
Key facts
- CB-APM embeds aggregate analyst consensus as a structural bottleneck
- Professional beliefs are treated as a sufficient statistic for market information
- Interpretability-by-design achieved through bottleneck constraint
- Bottleneck constraint functions as an endogenous regularizer
- Improves out-of-sample predictive accuracy
- Portfolios sorted on CB-APM forecasts show strong monotonic return gradient
- Robust across macroeconomic regimes
- Learned consensus encodes priced variation not spanned by canonical factor models
Entities
Institutions
- arXiv