The strategy · step by step

How it works

The rule of the game: keep what is rising, take shelter when everything falls — and the figures that prove it.

Portfolio Tester replays the past: it measures what an investment rule would have returned if you had followed it (a "backtest"). The strategy actually used splits the money into two pockets — 70% US equities, 30% commodities — and reviews its choices once a month.

The method is called dual momentum. It asks two questions each month. First: among a short list of investments, which is rising strongest right now? You hold that one. Then: is this winner truly rising, or is it just the least bad of a falling market? If it no longer beats a safe baseline, you sell and take shelter — in plain cash, that is, short-term US Treasury bills (BIL) — rather than riding the fall down. The shelter used to be a gold-and-cash blend; it is now cash: what protects in the crises that matter is stepping out cleanly, whereas a gold parachute is a bet on one particular market backdrop (see the 100-year proof).

Concretely: 70% of the capital rotates among three US equity funds (large growth companies, value companies, or the two blended), 30% among commodity sectors. Each pocket applies the rule on its own side; each month, the proportions are reset to 70/30. An investment's "strength" is measured over its last 1, 3 and 6 months, averaged.

Where dual momentum comes from

The older idea is relative momentum: among several investments, the winners of the last few months tend to keep winning over the next few — a fact documented by academic research as early as 1993 (Jegadeesh and Titman). On its own this reflex has a flaw: it always stays invested, and dutifully holds the "least bad" loser through a crash. Absolute momentum adds the missing gate: when the winner's own trend turns negative against a safe baseline, you step aside. Gary Antonacci combined the two in 2014 in his book Dual Momentum Investing: pick the strongest, but keep it only if it is truly rising. This site applies that rule each month, sleeve by sleeve, against a market barometer (SPY, the broad index of US equities, for the equity sleeve; DBC for commodities).

The proof on real instruments

The "100-year proof" page asks whether the logic survives across a century of market eras, on long stand-in series ("proxies"). Here is the complementary check: the actual US instruments over the window each one exists — the equity ETFs (exchange-traded funds) back to 2000, starting at the dot-com peak, an unflattering point; the commodity sectors back to about 1992; the composite to about 2001. The signal takes shelter in cash, never in a commodity barometer: the broad commodity index shown here is only a comparison line, not part of the rule. The two legs behave as the method predicts. This is a fidelity check, not a second claim of robustness. Below each equity curve, a panel shows how deep the declines run over time (the underwater drawdown).

And the two legs combined — the deployed 70/30, rebalanced monthly, not a simple overlay of the two curves above:

The real portfolio combines the two champions — US equity-style and commodity-sectors, weighted 70/30 — while deliberately leaving bonds out, a macro conviction rather than a performance one (see "Why these choices"). Combining the two did better than just add up: the worst decline of the whole is markedly shallower than either sleeve alone — that is the diversification the composite is built for. Over the 2016–2026 window the deployed 70/30 — the one that takes shelter in cash — makes about 16% a year with a worst decline near −10%: roughly the S&P 500's return over the same period at half the fall.

All figures are month-end closes; daily-marked drawdowns run deeper — see the note on granularity in the Guide.

These figures come from one historical window — the very one used to tune the strategy. They show it behaved well there; they do not prove the edge will survive other eras. A genuine test on data never used for tuning — including a true Monte-Carlo / bootstrap of returns — remains future work.

The shelter — what the strategy holds when it steps out of the market — is the real diversification lever, more even than the geography or style of the assets it rotates between. That shelter is now plain cash: in the crises that matter, the edge comes from stepping out cleanly, not from a clever safe-haven. Get the exit right and the whole portfolio behaves better.

How the universes are built

A "universe" is the short list of investments a sleeve picks from each month. Each list spans a single axis — US large-company style (growth / value / blend), geography (US / developed / emerging), commodity sectors (precious metals, energy, agriculture, base metals), or the quality of borrowers (governments / well-rated companies / riskier borrowers). The members of a list are chosen not to resemble each other: depending on the period, one or another leads — and that alternation is exactly what the rotation tries to capture. Each sleeve's shelter always comes from a different family of investments than the sleeve itself, so the refuge doesn't fall at the same time as what it replaces.

The crypto universes are the exception. Bitcoin alone is not a list — it is a single, extremely volatile asset — and the rule then collapses to an in-or-out decision: hold Bitcoin while its own trend beats cash, take shelter in cash otherwise. The Bitcoin/Ethereum pair adds a second choice and becomes a genuine rotation again. Both remain an exploratory, simulated testbed, never traded — the crypto section covers what they show and why their measurement window is shorter.

What each universe holds

Why US ETFs? The signals — momentum and the switch threshold — are computed on the dollar price series of US ETFs, which are the strategy's original reference assets and have a long history. In life-insurance wrappers you hold equivalent UCITS funds: they are for holding, not for the calculation, which stays on the US series. This is what guarantees the signal faithfully reproduces the backtested strategy.

In production the 70/30 only ever holds seven assets — IWF/IWD/IWB on the equity side, GLD/DBE/DBA/DBB on the commodity side — plus the cash refuge (BIL); SPY and DBC serve as read-only barometers, never held.

Finding the champions

Each family of investments was tested by replaying the 2016–2026 history under different settings — a series of targeted, hypothesis-driven trials (not an exhaustive grid, and not a Monte-Carlo simulation) — ranked by the return obtained per unit of worst decline endured: the CAGR (the average annual growth) divided by the maximum drawdown (the worst fall from a peak). The best setting in each family is its "champion."

Not every avenue we tested survived, and that is by design: the form offers only what held up. Rotating across the nine major US equity sectors (real estate and communications, too recent, were excluded) is an instructive case. With the same cash refuge as the deployed strategy, it contains risk remarkably well — a maximum drawdown of −11.5%, the shallowest of the entire equity class. But it returned only +9.4% a year over 2016–2026, six points below simply buying and holding the S&P 500 (+15.4%) — whereas the growth/value rotation we kept, with the same parachute, stays within a point of the index (+14.3%). Sector steering buys a safety the portfolio does not need to pay that much for: a great deal less drawdown, but six points of annual return given up. That is why sectors do not appear in the picker.

Defaults set by your choice of universe

The convictions behind the allocation

The allocation is set by deep convictions, not by recent performance: it is revised only when one of them is refuted — never because of a passing drawdown. Each one states what would prove it wrong.

Long bonds excluded. Long-maturity government debt is structurally unattractive. Refuted if the G7 debt-to-GDP ratio enters a durable decline, if 10-year US TIPS hold above 2% for 24 months, or if core inflation holds below 2% for 24 months.

Gold is insurance, not an engine. It protects in inflation and monetary crises and disappoints the rest of the time: a bet on one type of era. It therefore leaves the defensive sleeve (now plain cash) and keeps its place only as static catastrophe insurance, held outside the momentum system. Refuted if cash proves a worse defensive asset than gold over a full cycle.

Growth/value rotation (secular). The alternation reflects stable investor psychology, not a fashion; the rotation premium shows up across nearly every decade since 1920 (Fama-French's book-to-market criterion, see the century test), even though it deepens the drawdown in a crash with no style refuge, as in 1930. Refuted if the premium vanishes for 15 years.

Real assets and the AI build-out. AI lowers the price of services, not of real assets: data centres run on electricity and metals, which puts a structural bid under energy and materials. This is a worldview, not a hard-coded position — momentum takes it or leaves it. Refuted if AI capital spending contracts for 24 months, or if real-asset demand decouples from the growth of compute.

Last review: June 2026.