We explore the use of sentiment-driven demand, a key component of latent asset demand, in constructing mean-variance portfolios. Our approach decomposes these portfolios into an equally weighted component and an arbitrage component, and shrinks toward the equally weighted component and away from the arbitrage component when investor sentiment is low. This shrinkage technique helps to reduce estimation risk and imposes a tighter bound on the amount of asset mispricing that the arbitrage component can exploit when investor sentiment is low. Our results demonstrate the importance of considering latent demand in building robust investment strategies.
Lassance, N. (2022). Shrinking Against Sentiment: Exploiting Latent Asset Demand in Portfolio Selection. 16th International Conference on Computational and Financial Econometrics, London. https://hdl.handle.net/2078.5/106379