When considering long-short pairs trading candidates [background post], there are a number of important selection considerations.
One of these is minimization of trend component, which may be considered at both the constituent security/basket level, as well as within the pair ratio itself. One screen that has worked quite well for ETF Rewind subscribers, has been CSS Analytic’s* Livermore Trend Ranks.
In the table below, we assess the least trendy/ most mean reverting ranked securities (as reported late last month out of respect to subscribers), and conduct a brute force cross-pairs optimization test using the tools built into ETF Rewind, as follows:
As shown, a good number of the pairs performed quite nicely out of sample with frequent signals and good equity curve linearity. Of course, each selection needs to be tested for parameter robustness and bid-ask-volume feasibility before final selection. Nevertheless, one can easily imagine constructing robust baskets of either high probability up and down reverts, or more structured industry/ beta matched portfolios. ETFR + CSSA, what a terrific combination!
Related posts:
