We conduct inference on volatility with noisy high‐frequency data. We assume the observed transaction price follows a continuous‐time Itô‐semimartingale, contaminated by a discrete‐time moving‐average noise process associated with the arrival of trades. We estimate volatility, defined as the quadratic variation of the semimartingale, by maximizing the likelihood of a misspecified moving‐average model, with its order selected based on an information criterion. Our inference is uniformly valid over a large class of noise processes whose magnitude and dependence structure vary with sample size. We show that the convergence rate of our estimator dominates n 1/4 as noise vanishes, and is determined by the selected order of noise dependence when noise is sufficiently small. Our implementation guarantees positive estimates in finite samples.

We develop an asymptotic framework for conducting inference on continuous-time asset pricing models using high-frequency returns over an increasing time horizon. Our study focuses on the identification and estimation of risk premiums associated with the continuous component and jumps of various size brackets. We extend the classical Fama-MacBeth regression from the discrete-time setting to a continuous-time factor model, incorporating general dynamics for factors, idiosyncratic components, and factor loadings. Our empirical analysis of U.S. equities, foreign exchange, and commodities underscores the distinct significance of continuous and jump risk premiums for the specific factors constructed within each asset class in determining expected returns.

We propose a semiparametric two‐step inference procedure for a finite‐dimensional parameter based on moment conditions constructed from high‐frequency data. The population moment conditions take the form of temporally integrated functionals of state‐variable processes that include the latent stochastic volatility process of an asset. In the first step, we nonparametrically recover the volatility path from high‐frequency asset returns. The nonparametric volatility estimator is then used to form sample moment functions in the second‐step GMM estimation, which requires the correction of a high‐order nonlinearity bias from the first step. We show that the proposed estimator is consistent and asymptotically mixed Gaussian and propose a consistent estimator for the conditional asymptotic variance. We also construct a Bierens‐type consistent specification test. These infill asymptotic results are based on a novel empirical‐process‐type theory for general integrated functionals of noisy semimartingale processes.

Standard estimators of risk premia in linear asset pricing models are biased if some priced factors are omitted. We propose a three-pass method to estimate the risk premium of an observable factor, which is valid even when not all factors in the model are specified or observed. The risk premium of the observable factor can be identified regardless of the rotation of the other control factors if together they span the true factor space. Our approach uses principal components of test asset returns to recover the factor space and additional regressions to obtain the risk premium of the observed factor.

We perform a comparative analysis of machine learning methods for the canonical problem of empirical asset pricing: measuring asset risk premiums. We demonstrate large economic gains to investors using machine learning forecasts, in some cases doubling the performance of leading regression-based strategies from the literature. We identify the best-performing methods (trees and neural networks) and trace their predictive gains to allowing nonlinear predictor interactions missed by other methods. All methods agree on the same set of dominant predictive signals, a set that includes variations on momentum, liquidity, and volatility.

Data snooping is a major concern in empirical asset pricing. We develop a new framework to rigorously perform multiple hypothesis testing in linear asset pricing models, while limiting the occurrence of false positive results typically associated with data snooping. By exploiting a variety of machine learning techniques, our multiple-testing procedure is robust to omitted factors and missing data. We also prove its asymptotic validity when the number of tests is large relative to the sample size, as in many finance applications. To improve the finite sample performance, we also provide a wild-bootstrap procedure for inference and prove its validity in this setting. Finally, we illustrate the empirical relevance in the context of hedge fund performance evaluation.

We show that two important issues in empirical asset pricing—the presence of weak factors and the selection of test assets—are deeply connected. Since weak factors are those to which test assets have limited exposure, an appropriate selection of test assets can improve the strength of factors. Building on this insight, we introduce supervised principal component analysis (SPCA), a methodology that iterates supervised selection, principal‐component estimation, and factor projection. It enables risk premia estimation and factor model diagnosis even when weak factors are present and not all factors are observed. We establish SPCA's asymptotic properties and showcase its empirical applications.

We reconsider trend‐based predictability by employing flexible learning methods to identify price patterns that are highly predictive of returns, as opposed to testing predefined patterns like momentum or reversal. Our predictor data are stock‐level price charts, allowing us to extract the most predictive price patterns using machine learning image analysis techniques. These patterns differ significantly from commonly analyzed trend signals, yield more accurate return predictions, enable more profitable investment strategies, and demonstrate robustness across specifications. Remarkably, they exhibit context independence, as short‐term patterns perform well on longer time scales, and patterns learned from U.S. stocks prove effective in international markets.

We propose a model selection method to systematically evaluate the contribution to asset pricing of any new factor, above and beyond what a high‐dimensional set of existing factors explains. Our methodology accounts for model selection mistakes that produce a bias due to omitted variables, unlike standard approaches that assume perfect variable selection. We apply our procedure to a set of factors recently discovered in the literature. While most of these new factors are shown to be redundant relative to the existing factors, a few have statistically significant explanatory power beyond the hundreds of factors proposed in the past.

We propose an approach to measuring the state of the economy via textual analysis of business news. From the full text of 800,000 Wall Street Journal articles for 1984 to 2017, we estimate a topic model that summarizes business news into interpretable topical themes and quantifies the proportion of news attention allocated to each theme over time. News attention closely tracks a wide range of economic activities and can forecast aggregate stock market returns. A text‐augmented vector autoregression demonstrates the large incremental role of news text in forecasting macroeconomic dynamics. We retrieve the narratives that underlie these improvements in market and business cycle forecasts.