quantlab.risk.predefined.use4#

USE4-style factor risk model: the factor set, the regression and the estimate chain.

Follows Menchero, Orr and Wang, The Barra US Equity Model (USE4), Methodology Notes (MSCI, 2011), §3-5 and Appendix B: one country factor, one factor per industry (the point-in-time Fama-French 48 industries) and the 12 USE4 style factors, 61 factors, by default on the BarraStyle outputs; any factor’s outputs can stand in through Use4RiskConfig. Its regression store holds, for every bar t, the cross-sectional regression of the excess returns of t on the exposures of t-1:

  • the excess return of a symbol is its adjusted close of t over that of t-1, less one, less the risk-free rate of t-1 (the rate of a day is published the next business day; BarraStyle lags it the same way);

  • exposures: 1 on the country factor, 1 on the symbol’s industry and 0 on every other industry, and its style exposures;

  • weighted least squares over the estimation universe of t-1 (every symbol without one), weighted by config.weighting of the market cap of t-1; with both a country factor and industries, subject to the cap-weighted industry factor returns summing to 0, which removes their collinearity and makes the country factor the cap-weighted market (USE4 Methodology Notes eq. 3.3);

  • an industry with fewer than min_industry_members fitted members is left out of the bar: it has no factor return that bar, its members are not in the fit, and their specific returns carry no industry term (our choice);

  • a fitted return more than return_outlier_sigma robust standard deviations (1.4826 times the median absolute deviation) from the cross-sectional median is trimmed to that bound for the fit only. A robust bound does not move when the outlier grows, so a vendor price error cannot move a factor return (our choice);

  • the specific return of every symbol with all exposures and a return, fitted or not, is its untrimmed excess return less the fitted factor part.

Classes

Use4RiskModel

USE4-style factor risk model: country, industries and styles; EWMA, Newey-West, eigenfactors, VRA.