Weak Signals, Small Bets: A Portfolio Perspective on Firm Characteristics

Abstract

We examine how the choice between sparse variable selection and continuous shrinkage affects portfolio performance when many noisy stock characteristics compete for allocation. Ridge-type regularization economically dominates sparsity-inducing approaches after transaction costs. Sparse methods concentrate weights on fewer characteristics, generating costly turnover; shrinkage distributes exposure broadly, producing diversified portfolios that are cheaper to rebalance. Aggregate transaction costs provide natural guidance for balancing this trade-off. Our results hold across samples, market conditions, implementation constraints, and alternative Bayesian priors, suggesting that the choice of approach to mitigate overfitting is fundamentally an economic decision, not merely a statistical one.

Download

• Working Paper

Awards

REAG Investimentos Best Articles Award, XXV Brazilian Finance Meeting at INSPER.

Presentations

Alpine Finance Summit (scheduled), Barcelona Workshop in Financial Econometrics, 11th Italian Congress of Econometrics and Empirical Economics, XXV Brazilian Finance Meeting at INSPER.

Citation

Bianchi, Daniele, and Pedro H. M. Venturi. “Weak Signals, Small Bets: A Portfolio Perspective on Firm Characteristics.” Working paper.

@article{bianchi2024weak,
  title={Weak Signals, Small Bets: A Portfolio Perspective on Firm Characteristics},
  author={Bianchi, Daniele and Venturi, Pedro},
  journal={Small Bets: A Portfolio Perspective on Firm Characteristics (July 27, 2024)},
  year={2024}
}