Evaluating an annual equity-gold momentum rank rotation

One yearly procedure

The evaluated rule is a yearly momentum-strategy: compare trailing annual total returns of a broad equity index and gold, then hold the leader through the next calendar year. In the fixed terminology, a momentum-strategy is a rule that uses trailing relative strength, here last calendar year's total return, to choose the next holding period's asset.

The same rule is a two-asset rank-rotation: at each year-end the two markets are ordered by that annual return and only the top-ranked name is held for the system holding period. Rank-rotation is a fixed-date ordering of the eligible assets that assigns the system holding period to the higher-ranked name.

The one-year hold was fixed from a mix of measured system behavior and tax-year practicality, not from intra-year rebalancing. That annual-hold is a one-year, year-end-to-year-end window that is both the signal horizon and the rebalance calendar.

The winner-weight sweep

Stability was tested with a winner-weight K: K percent of capital goes to the trailing leader and the rest to the laggard, with K swept from 100 percent down to 0 percent. Winner-weight is the share of capital placed in the trailing leader; the complement stays in the laggard.

In that sweep, K of 50 percent or more was scored as overweighting the leader versus an equity-index baseline, and the highest recorded terminal value of a one-unit stake occurred at K equal to 100 percent.

TradersWeek editorial: the sweep is one of the two grades applied to the locked procedure. It asks whether changing the winner-weight changes how that annual rule should be read.

Case windows and a post-stress interval

Calendar windows around 2000-2002 and 2008 were used as case studies of whether the rank rule had already rotated into gold after gold led in the prior year.

The same yearly rank procedure was also scored over 2009-2018 as a post-stress interval after a long equity advance.

Research-series versus implementation-vehicle

The long evaluation window runs from 1971 through 2018 and begins after official currency-to-gold convertibility ended, so the sample is framed as a post-convertibility market.

That long sample is built from an equity index series and gold spot prices because corresponding exchange-traded fund histories were too short for the full window. Those non-fund histories are the research-series.

An implementation check of the same annual rank rule on index-or-spot data versus exchange-traded fund data over 2004-2018 reported a 0.998 correlation between the two result series. The exchange-traded fund wrapper is the implementation-vehicle compared with the research-series over the years both exist.

TradersWeek editorial: this check is the second grade. It asks whether reading the procedure on the research-series changes once the implementation-vehicle is available.