Invictus Construction

Episode five. Construction. The educational book is a nineteen-holding, four-sleeve architecture designed to compound across twenty to thirty years without requiring a correct forecast. Ray, the design intent in one paragraph.

Most portfolios are overexposed to growth because equities dominate both capital and risk. All Weather seeks roughly equal risk across the four growth-inflation environments. Growth core is the wealth engine. Real assets protect purchasing power when inflation surprises higher. Defensive limits permanent impairment in contraction. Thematic is small enough that thesis failure is survivable and large enough that secular success matters. No single macro outcome is required for solvency. That is the institutional standard.

I own businesses, not sleeves. I will still say this: if you are going to use funds, know what they own. A global equity accumulator is a claim on corporate earnings. An S-and-P accumulator is a United States concentration, not diversification. Emerging markets at moderate correlation are a different animal. Minimum volatility is still equity beta with the left tail trimmed, not a bond.

Say the quiet parts. Gold has no yield. It is a monetary asset driven by real rates, currency confidence, and scarcity. Listed infrastructure has partially inflation-linked cash flows and is still equity. Commodity products earn spot, collateral, and roll — and contango can be a persistent drag. A listed global-macro closed-end vehicle is manager-dependent and trades at a premium or discount. Never describe it as a guaranteed crisis hedge. Its value is the chance of low equity correlation when the process works.

Two redundancies were the most damaging. A global equity accumulator against an S-and-P accumulator at about point nine seven. A global aggregate against a long Treasury sleeve at about point nine five. Those are duplicates wearing different tickers. After clean-up, pairs like robotics against industrial metals near zero, and gold against broad commodities around point four, are actual diversification. Correlation is regime-dependent. In two thousand eight, listed risk sold together. Physical gold and a true macro process can behave differently. That difference is the only reason to pay for the extra engine.

Currency is the amateur trap. Listing currency is not economic exposure. A sterling-listed ETF can carry substantial dollar, euro, and emerging-market currency risk if the underlying is unhedged. Look-through currency attribution is mandatory. So is a single synchronised return matrix. Do not average constituent metrics. Compute portfolio return, then NAV, then volatility, Sharpe, drawdown, and risk contribution on that one series.

Risk contribution. Give the arithmetic so it cannot be waved away.

Portfolio variance is not the weighted average of variances. Covariance terms dominate. A one-percent thematic with thirty-five percent volatility and point-seven correlation to the growth core can add more to total volatility than a five-percent low-vol bond sleeve. Capital size does not equal risk size. Satellites must be examined on marginal contribution to volatility and to expected shortfall under historical archetypes. The maximum size of an expression is the residual risk budget after the core is funded, reduced by hedge effectiveness, conviction in the gap, liquidity, and operations. No idea, however elegant, is allowed to permanently impair the compounding engine.

There is also volatility drag. The expected compound growth rate is approximately the arithmetic mean minus half the variance. High-vol assets need higher arithmetic returns just to match the terminal wealth of lower-vol assets. That is one reason genuine diversifiers and minimum-volatility tilts can improve terminal wealth even when their arithmetic means look modest.

The contribution schedule is the unsung hero. A fixed monthly amount on a calendar date buys more units when prices fall and fewer when they rise. Rebalancing sells relative winners and buys relative losers. Both are mechanical. Both remove the behaviour that has destroyed more retail compound returns than any factor: the decision to pause when it feels terrible and to add when it feels clever.

History is the laboratory, not the crystal ball. Nineteen twenty-nine to thirty-three: deflationary deleveraging, cash and later quality duration. Nineteen-seventies: stagflation, gold and real assets. Japan after nineteen-ninety: balance-sheet recession, bonds and cash for a long time. Two thousand eight: funding shock, quality duration and cash, listed alternatives stressed. Two thousand twenty-two: inflation shock, stocks and bonds down together, gold and commodities differentiated. If your stated diversifier failed in the analogue that matches its job, it is not a diversifier. It is a mascot.

The supplied case-study window from February twenty twenty-four to August twenty twenty-six showed an attractive compound rate with mid-single-digit volatility and a drawdown near ten percent. Treat those as an educational reconstruction, not an audited live track. Short windows flatter recent gold, defence, and uranium. Forward assumptions must be more conservative. Principle of episode five. Structure first, time second, prediction last. Next: Bloomberg as the measurement system, desk language, and the closing principle that ties the household, the ticket, and the book to the same machine.