Claim
Frank’s heuristic, from 30 years of trading (recorded faithfully; untested): gold has call-option-like convexity — it bleeds a portfolio through its lost-decade chop (1980–2000), but the bleed “doesn’t hurt all that bad if you add the momentum filter and add the position with leverage.” I.e. trend-filter gold to suppress the chop years, and size the filtered exposure levered, and the option-like payoff profile survives net of the bleed.
origin: frank— recorded for the record, to be independently re-researched.- Why it might be true: the filter truncates the premium-payments phase (the 1980–2000 bleed is almost entirely below the 200-month MA / negative 12-month trend); leverage on the filtered (trend-positive) subset is how optionality gets expressed in a linear-instrument portfolio. Related vendor evidence cuts both ways: gold-trend-filter-underperforms-bnh says the filter costs return vs gold B&H — but says nothing about risk-adjusted or levered comparison, which is the claim’s actual ground.
- Why to be skeptical: whipsaw costs in choppy-but-rising stretches (2003–06 type regimes) plus financing cost on the leverage can eat the convexity; the claim’s truth depends heavily on the financing rate and filter parameters — testable but parameter-sensitive.
- Test (to propose): gold trend-filtered, 1971+, at leverage 1×/2×/3× vs a bill-financed and margin-financed cost model; compare CAGR, max DD, and worst-5-year rolling return vs gold B&H and vs T-bills. Key stat: does the filter+leverage combo ever produce a rolling 5-year loss worse than B&H’s?