The Chen-Tindall system and the lasso operator: improving automatic model performance
Using U.S. monthly macroeconomic data, the automatic model system presented in Chen and Tindall  outperforms the lasso automatic system, but the lasso is improved where Bayesian model averaging is employed to combine its forecasts with those from autoregressive schemes. The best performance is obtained using Bayesian model averaging to combine the Chen?Tindall system, the lasso, and autoregressive schemes. Performance is virtually the same using this combined approach where the elastic-net operator is substituted for the lasso. Similar overall outcomes are found for France and Germany ...
Understanding hedge fund alpha using improved replication methodologies
In this paper, we estimate alpha for major hedge fund indexes. To set the stage, we examine several alternative methods for replicating Hedge Fund Research Inc. hedge fund indexes. The replication methods include stepwise regression, variations of the lasso shrinkage method, principal component regression, partial least squares regression, and dynamic linear regression. We find that the lasso methods and dynamic regression are superior for generating hedge fund replications and that the performance of the replications corresponds closely to that of the respective actual indexes. Using these ...
Treasury Auctions During the Pandemic: Stresses but Few Surprises
The Treasury auction surprise indicator suggests that, despite a sharply negative reading in March 2020, Treasury auction outcomes have normalized.
Risk measurement illiquidity distortions
We examine the effects of smoothed hedge fund returns on standard deviation, skewness, and kurtosis of return and on correlation of returns and cross-sectional volatility and covariance of returns using an MA(2)-GARCH(1,1)-skewed-t representation of returns instead of the traditional MA(2) model employed in the literature. We present evidence that our proposed representation is more consistent with the behavior of hedge fund returns and that the traditional method tends to overstate the degree of smoothing observed in hedge fund returns. We present methods for correcting for the distortive ...
The structure of a machine-built forecasting system
This paper describes the structure of a rule-based econometric forecasting system designed to produce multi-equation econometric models. The paper describes the functioning of a working system which builds the econometric forecasting equation for each series submitted and produces forecasts of the series. The system employs information criteria and cross validation in the equation building process, and it uses Bayesian model averaging to combine forecasts of individual series. The system outperforms standard benchmarks for a variety of national economic datasets.
Constructing Zero-Beta VIX Portfolios with Dynamic CAPM
This paper focuses on actively managed portfolios of VIX derivatives constructed to reduce portfolio correlation with the equity market. We find that the best results are obtained using Kalman filter-based dynamic CAPM. The portfolio construction method is capable of constructing zero-beta portfolios with positive alpha.
Hedge fund dynamic market sensitivity
Many hedge funds attempt to achieve high returns by employing leverage. However, it is difficult to track the degree of leverage used by hedge funds over time because detailed timely information about their positions in asset markets is generally unavailable. This paper discusses how to combine shrinkage variable selection methods with dynamic regression to compute and track hedge fund leverage on a time-varying basis. We argue that our methodology measures leverage as well as hedge fund sensitivity to markets arising from other sources. Our approach employs the lasso variable selection ...
Hedge Fund Return Prediction and Fund Selection: A Machine-Learning Approach
A machine-learning approach is employed to forecast hedge fund returns and perform individual hedge fund selection within major hedge fund style categories. Hedge fund selection is treated as a cross-sectional supervised learning process based on direct forecasts of future returns. The inputs to the machine-learning models are observed hedge fund characteristics. Various learning processes including the lasso, random forest methods, gradient boosting methods, and deep neural networks are applied to predict fund performance. They all outperform the corresponding style index as well as a ...
Volatility-selling strategies carry potential systemic cost
Investors have increasingly turned to stock market volatility-selling strategies based on the idea of selling implied volatility and buying it back later when it falls to a level more consistent with realized volatility.
Dynamic Methods for Analyzing Hedge-Fund Performance: A Note Using Texas Energy-Related Funds
We apply dynamic regression to Texas energy-related hedge funds to track changes in portfolio structure and manager performance in response to changing oil prices. We apply hidden Markov models to compute shifts in portfolio performance from boom to bust states. Using these dynamic methods, we find that, in the recent oil-price decline, these funds raised their exposure to high-grade energy-related bonds in a bet that the spread to low-grade energy bonds would widen. When the high-grade bonds eventually fell, the hedge funds entered into a bust state.