Fahiz Baba‑Yara

The multifactor risk-return tradeoff turns strongly positive once factor covariances, not just variances, are accounted for. Covariances explain more variation in factor returns than variances and matter for both in-sample and out-of-sample prediction.

Employee news consumption helps characterize firms' exposure to macroeconomic risk. Firms whose employees read more macroeconomic news in advance are more exposed when conditions worsen later. Those firms also hedge more, face higher costs of capital, and invest and hire less.

Earlier public records also appear under the title Are Uncertain Firms Riskier? with a different coauthor line. This entry follows the March 2024 draft title and author line.

Firm characteristics contain both persistent and transitory components, and the pricing of characteristics is driven largely by the persistent component. Strategies built on the persistent component deliver much higher Sharpe ratios than standard characteristic-based strategies.

Around four-fifths of commodity futures strategies that were profitable before financialization stop being profitable afterward. The decline appears to come from adverse repricing of a few systematically important factors.

Value spreads predict value returns in equities, industries, commodities, currencies, and global bond and stock markets. A single common component explains about two-thirds of that predictability and aligns with broad variation in risk premia.

Modern machine-learning factor models achieve high Sharpe ratios in very different ways and disagree on which factors are actually priced. Much of the return attributed to factor exposure by one model is treated as alpha by another.

Unconstrained machine-learning models can predict short-run cross-sectional returns without generalizing to the time-series behavior of the market and long-short portfolios. Imposing economically motivated structure restores that out-of-domain generalization.