Evaluating momentum rank rotation under portfolio constraints

From comparison to a portfolio procedure

Relative strength was framed as comparing one item's results with another over a chosen window. The evaluation asked whether that ranking can be turned into a portfolio procedure rather than left as a broad screen or an extreme-tile sort.

The tested universe was compared with an equal-weighted index of that same universe. Two public momentum rankings were converted to percentiles: trailing return over a chosen window, and closing-price deviation from a simple moving average.

Buy ranks, sell ranks and replacement

The book was run as a rank-rotation system with a target name count, a percentile buy threshold, a lower percentile sell threshold, weekly review and replacement only after a holding crossed the sell gate.

The percentile buy threshold is the minimum relative-strength percentile that makes a name eligible for entry. The percentile sell threshold is the relative-strength percentile that forces an automatic exit and replacement.

Random draws from the eligible basket

Because buying an entire top slice of the ranked list was treated as impractical, each purchase drew one name at random from the eligible basket. Monte Carlo simulations were run for every parameter set.

Holding caps and a fully invested book

A holding was trimmed if it exceeded a set share of portfolio value. The book stayed fully invested, and a new buy was allowed only after a corresponding sale.

Practicality checks and turnover

Before the book was run, practicality checks required a universe broad enough for different objectives, return dispersion that lets a smaller name subset keep similar characteristics, year-to-year consistency rather than a single-year result, and robustness across both formulations. Return dispersion is the spread of individual outcomes inside a ranked basket that can keep a small subset from matching the basket average.

Estimated turnover for the trailing-return rotation was described as comparable to an average given for actively managed domestic stock funds. The moving-average rotation produced much higher turnover and was framed as suitable only for very active implementation after costs and taxes.