Does it make sense to use sectors with the Dual Momentum model? In Gary Antonacci’s book, he suggests the following – “My favorite dual momentum strategy is one that rotates among the strongest U.S. stock market equity sectors.” (page 131-133) In this blog post I examine this approach using Fidelity sector ETFs. The Fidelity real estate ETF (FREL) is new so I included VNQ from Vanguard. To provide additional information I also included VTI and BND. Keep in mind that the Fidelity ETFs do not have a three-year record so this information uses the past 26 months for analysis.
Sector Rankings: Based on 1/20/2015 data, SHY is the top ranked ETF followed by BND, Vanguard’s bond ETF. I presume Antonacci would invest 100% in the bond fund since no sector ETF is performing above SHY. That is a guess as I don’t have information as to how he manages sector rotations.
Sector Clusters: Cluster analysis provides more information as back-tests point out the importance of finding low correlated securities. One of the reasons I included VTI in this analysis was to see how many sectors were highly correlated with VTI. Six of the eleven Fidelity sectors have a correlation of 0.8 or higher. We don’t gain much by including these highly correlated sectors when constructing a portfolio. We expect FREL and VNQ to be highly correlated as both are real estate ETFs.
Those sectors that could add potential value are FUTY, FENY or FMAT, and FHCL, as you can see from the cluster analysis below.
My next move was to mix the “Rutherford 10” with the Fidelity sectors to see what sectors carry a low correlation with ETFs frequently used to populate our ITA portfolios.
Expanded Cluster Analysis: Here we have a cluster analysis that is a mixture of ETFs used to populate most ITA portfolios plus the Fidelity sector ETFs. Based on current data, sectors of interest are: Healthcare (FHCL), Utilities (FUTY), Energy (FENY) and Materials (FMAT). Energy may be an anomaly due to the current crash in oil prices. Further, I would want to run this same analysis using Vanguard sectors in order to capture 36 months of data.
We might gain a little by adding a few sectors to our “Rutherford 10” or “Baker’s Dozen,” but I doubt the additions will add much alpha to our portfolios.
I made the error of introducing VNQ twice, so mentally eliminate one.