Research

Portfolio Choice with Target Horizon (Job Market Paper)

Presentations: Financial Management Association Annual Meeting (2025), University of Arizona (2025)

Abstract: I propose a new approach to multi-period portfolio choice in high-dimensional settings that considers important aspects of the joint distribution of holding-period returns while being agnostic about the dynamics of investment opportunities. I characterize optimal final payoffs through factor principal components of holding-period returns and form tradable factor-tracking portfolios that deliver these horizon-specific payoffs. The resulting optimal portfolio is a utility-maximizing combination of the factor-tracking portfolios. Using broad samples of equity portfolios as investable test assets, I show economically large utility gains for investors compared to important benchmarks, both in-sample and out-of-sample. More importantly, the estimated portfolio rules are horizon-specific: optimal short-horizon strategies cost investors with long horizons, while optimal long-horizon strategies appear deeply unattractive when evaluated by their short-term returns. The findings provide compelling evidence for the economic value of investment horizon in portfolio decisions and challenge the conventional focus on short-term performance metrics for evaluating portfolios of long-horizon investors.

Shrinking the [Efficient] Cross-Section (Work in Progress)

Presentations: University of Arizona (2023, 2024)

Abstract: I apply factor-timing approaches to construct "efficient" versions of anomaly portfolios. Unlike the original cross-section, where up to 50 principal components (PCs) contribute to out-of-sample Sharpe ratios, the efficient space exhibits sparsity: only 5-10 PCs matter, achieving nearly double the performance. This suggests that factor proliferation in academic studies reflects systematic noise rather than distinct economic signal. After removing predictable patterns through factor-timing, which sophisticated arbitrageurs would naturally exploit, only factors driving common variation are rewarded, reconciling empirical findings with theoretical predictions about arbitrage-limited markets.

Time-Series Efficient Factors and the Cross-Section of Returns (Working Paper)

Presentations: University of Arizona (2022, 2023)

Abstract: A practical mean-timing approach systematically enhances asset pricing models and generates higher Sharpe ratios across a broad range of assets. Evaluating 206 anomaly portfolios, I document a significant improvement in models that incorporate time-series efficient (TSE) factors for pricing returns. On average, models using TSE factors correctly price 17.6% more anomaly portfolios than original models. This improvement extends beyond renowned asset pricing models; randomly constructed three-factor models using TSE factors outperform their counterparts that employ original factors. Notably, time-series efficient trading strategies exhibit an average 21.7% increase in out-of-sample Sharpe ratios, underscoring their potential value for real-time investors.

How AI Affects the Future Relationship Between Corporate Governance and Financial Markets: A Note on Impact Capitalism (Published)

Managerial Finance (2022, with R. McBride and A. Dastan)

Abstract: We discuss how artificial intelligence (AI) can facilitate and incentivize reform in corporate social responsibility (CSR), i.e. governance with regard to pledge for socially responsible investments (SRIs). In a closed system, there is a reinforcing feedback loop between SRI and CSR. AI is a moderator to increase SRI and a mediator to incentivize CSR. If the legal and ethical provisions are involved in the AI systems, they could act as catalyst for corporate governance reform towards sustainability.