Assistant Professor of Finance · Kelley School of Business · Indiana University
Fahiz Baba‑Yara
Empirical asset pricing is a measurement problem pretending to be a discovery problem. I build tools to separate persistent characteristics from transitory noise, then test what actually predicts returns.
I’m an empirical asset pricing researcher at Indiana University’s Kelley School of Business. I study return predictability, factor models, and what machine learning does once you make it respect basic asset‑pricing discipline.
Research interests: empirical asset pricing, return predictability, factor portfolios, and machine learning in finance.
- Signal extraction: persistence vs. noise
- Cross‑asset return predictability
- ML factor models & robustness
Selected work
Full listResearch
Selected publications and working papers. Click a title for a short abstract, a one‑line takeaway, and a BibTeX citation you can copy.
2024
Persistent and Transitory Components of Firm Characteristics
with Martijn Boons &
Andrea Tamoni
Journal of Financial Economics
Persistent and Transitory Components of Firm Characteristics
Abstract
We decompose firm characteristics into persistent and transitory components. We find that the pricing of characteristics is almost exclusively driven by their persistent components. A strategy that trades on the persistent component yields Sharpe ratios that are more than double those of standard characteristic-based strategies.
2021
Value Return Predictability Across Asset Classes
with Martijn Boons &
Andrea Tamoni
Review of Finance
Value Return Predictability Across Asset Classes
Abstract
We show that returns to value strategies in individual equities, industries, commodities, currencies, global government bonds, and global stock indexes are predictable in the time series by their respective value spreads. A single common component captures about two-thirds of value return predictability.
2021
The Limits of Factor Model Spanning
with Brian H. Boyer &
Carter Davis
R&R · Review of Financial Studies
The Limits of Factor Model Spanning
Abstract
We investigate the extent to which modern academic machine learning models agree on which factors are priced in the cross-section of returns. We find considerable disagreement across models that achieve high Sharpe ratios in remarkably different ways. The majority of average returns attributed to factor exposure by any given model is generally deemed pure alpha by other models. We develop a theoretical model with complexity that suggests machine learning models using available data are unlikely to ever span others, although they achieve high Sharpe ratios and produce additional novel insights.
2025
The Multifactor Risk-Return Tradeoff
with Martijn Boons &
Rik Frehen
Working paper
The Multifactor Risk-Return Tradeoff
Abstract
The multifactor risk-return tradeoff is severely understudied relative to the market. In contrast to mixed evidence for the market, we find that the multifactor risk-return tradeoff is strongly positive when appropriately accounting for factor covariances. Our multifactor risk model shows that covariances contribute more to variation in factor returns than variances and performs at least as well in predicting multiple factors as benchmark models specifically designed to predict a single factor, both in- and out-of-sample. Consistent with a positive multifactor risk-return tradeoff, conditional multifactor alphas for a large set of anomalies are indistinguishable from unconditional alphas.
2025
Risk from the Inside Out: Employee News Consumption
with Fotis Grigoris &
Preetesh Kantak
Working paper
Risk from the Inside Out: Employee News Consumption
Abstract
We propose a novel measure of firm risk based on the news consumption patterns of employees inside the firm. We show that when employees read more negative news about the economy or their sector, the firm's future stock returns are significantly lower.
2022
Commodity Returns: Lost in Financialization
with Massimiliano Bondatti
Working paper
Commodity Returns: Lost in Financialization
Abstract
We study the effect of the growth in investment capital on the average returns of popular commodity futures trading strategies over time. We find that approximately 80% of commodity futures strategies that generated statistically significant average returns before financialization (2004) are no longer profitable. The decline in strategy returns is primarily driven by an adverse change in the average returns of a few systematically priced factors in the cross-section of commodity futures. Commodity strategies with relatively higher exposure to the Dow Jones Commodity Index experience a significant reduction in average returns.
2020
Machine Learning and Return Predictability Across Firms, Time and Portfolios
Solo-authored
Working paper
Machine Learning and Return Predictability Across Firms, Time and Portfolios
Abstract
Previous research finds that machine learning methods predict short-term return variation in the cross-section of stocks, even when these methods do not impose strict economic restrictions. However, without such restrictions, the models' predictions fail to generalize in important ways, such as predicting time-series variation in returns to the market portfolio and long-short characteristic sorted portfolios. I show that this shortfall can be remedied by imposing restrictions that reflect findings in the financial economics literature in the architectural design of a neural network model and provide recommendations for using machine learning methods in asset pricing. Additionally, I study return predictability over multiple future horizons, shedding light on the dynamics of intermediate and long-run conditional expected returns.
Teaching
Contact
Email is best. If you’re writing about a paper, a replication, or a data question, include a one‑paragraph description and links to anything you want me to look at.
1309 E. Tenth Street
Bloomington, IN 47405