US and A Factor Momentum

We implement a forward looking long-only factor momentum approach, which attempts to time the market by exploiting predictability in common equity factors, including Fama French factors (i.e. market beta, SMB, HML, CMA, RMW, MOM and STRV), as well as factors put forth by AQR (i.e. DHML, BAB & QMJ). Factor momentum is thereby determined by running weighted least squares regressions on past monthly factor returns (individually for each factor), in order to implement an autoregressive process of order one (i.e. WLS-AR1). Since we focus on the long-leg of factor portfolios, we only choose factors which exhibit a positive predicted return over the next month (as indicated by WLS-AR1). The corresponding investments are then derived by computing the loadings of stocks within the Russell 3000 equity index (i.e. US stocks only) to the factors identified in the prior step (i.e. factors which are predicted to exhibit positive factor momentum). Using this top-down stock selection process, we then sort all stocks within the Russell 3000 by their factor loadings and finally conduct investments in the 100 stock (exact number may vary slightly), which exhibit the highest loadings and are thus most likely to correspond to the long-legs’ factor portfolios.