Personal Project · Strategic Asset Allocation
A 20–30 year strategic asset allocation derived from a single macro base case — that policy rates stay below inflation for an extended period — and carried through to a final allocation that holds no fixed income at all.
If nominal bonds and cash no longer behave as safe assets, what does a multi-decade portfolio actually look like?
Most retail allocation advice starts from a 60/40 split and adjusts at the margin. I wanted to start one layer earlier — from an explicit view about the monetary regime — and let that view decide every weight, including the weights it forces to zero.
The working assumption: governments and central banks hold policy rates below inflation to erode the real value of sovereign debt. Under that base case, nominal cash and nominal bonds have a negative expected real return — they stop being the risk-free anchor a classical allocation treats them as.
Every asset class is then judged on one question: does it help or hurt — on growth, inflation-hedging, diversification, or liquidity? That single test disqualifies nominal bonds, excess cash and illiquid alternatives together, rather than as a series of separate late objections.
As a retail investor with no formal liability schedule, the honest move was to say so explicitly rather than invent one. The portfolio follows a strategic total-return approach with a dedicated liquidity reserve, not a fully liability-matching mandate — but it borrows the LDI habit of separating a predictable liquidity portfolio from a growth portfolio that isn't meant to pay next year's bills.
Strategic asset allocation is the entire exercise. Tactical allocation is explicitly rejected — it requires market-timing skill the rest of the portfolio's logic argues against.
Broad ETF cores, with single-name satellites admitted only where they pass a "what diversification disappears if I drop this" test against the core's actual holdings.
Classical risk parity levers bonds so their low volatility pulls its weight — which fails badly in a simultaneous rate-and-equity shock, as 2022 showed. The leverage is dropped, not the diversification idea.
Anchor to market-implied equilibrium returns, then overlay explicit named macro views. Mean-variance optimization was acknowledged but not formally modeled — its input sensitivity doesn't suit a 20–30 year retail portfolio.
Position sizes are proportional to conviction rather than equal-weighted. The zeros carry as much of the thesis as the positives do.
The growth engine — pricing power is the only lever that can outrun suppressed policy rates over decades, which no bond can do. Split across Quality, Value and Momentum factors at 50/25/25 rather than a single factor bet, with emerging markets at a quarter of the sleeve and kept cap-weighted.
Listed, globally diversified core plus two satellites. Weighted above infrastructure for its stronger current yield and broader core index.
Regulated, often inflation-linked cash flows. The one satellite here exists purely to fill a sector gap the core index leaves at roughly zero.
The purest repression hedge in the portfolio — the one sleeve whose return mechanism is structurally independent of the equity engine. Held through a vehicle chosen for German tax treatment and physical delivery optionality rather than on cost alone.
A liquidity anchor, not a return position — and a 5% sleeve does not mean 5% a year gets spent. The draw order is portfolio income first, then this reserve, then systematic rebalancing out of whatever is overweight.
Excluded. Nominal bonds fail the macro thesis; the illiquid alternatives fail the liquidity framework. Commodities were eliminated after the only accessible vehicle turned out to be synthetic, exposed to roll yield, and — given its positive equity correlation — not actually a repression hedge at all.
An early review of this work landed a fair criticism: the weights had been reasoned qualitatively without ever stating a numeric input. These are standard long-run planning assumptions — directionally reasonable, not decimal-precise, and not a fitted model output.
| Asset Class | Expected Return | Expected Volatility | Rough Sharpe |
|---|---|---|---|
| Public Equity (DM factors) | ~7.5% | ~15% | ~0.33 |
| Emerging Market Equity | ~8% | ~20% | ~0.28 |
| Real Estate (listed) | ~7% | ~18% | ~0.25 |
| Infrastructure (listed) | ~7% | ~14% | ~0.32 |
| Gold | ~5% | ~15% | ~0.17 |
| Money Market | ~2.5% | ~1% | risk-free proxy |
The useful conclusion is a negative one: expected Sharpe ratios cluster closely across equity and infrastructure, so the marginal case for pushing equity beyond 60% is weak on risk-adjusted grounds alone. The 60% comes from combining that bound with the qualitative conviction that equity is the only sleeve whose pricing power can outrun repression over decades — not from the Sharpe ratios dictating a unique optimum. Anyone reviewing this should expect exactly that: rough quantitative bounds plus stated conviction, not a claim of precision the inputs can't support.
The portfolio is built for one regime. The more useful exercise is naming what happens in the other four.
| Scenario | Winners | Losers | Portfolio Impact |
|---|---|---|---|
| Financial repression (base case) | Gold, Infrastructure, Real Estate | Fixed Income (already 0%) | Performs as designed |
| Disinflation, rates normalize | Bonds, high-quality duration | Gold | The largest vulnerability. 0% fixed income means missing the one scenario where bonds would have outperformed, while gold's opportunity cost rises as real rates normalize. |
| AI / productivity boom | Equities, especially Quality and Momentum | Gold | The 60% equity sleeve still captures most of this upside; gold is a modest drag, not a large one |
| Recession / growth scare | Cash, high-quality bonds | Cyclical, housing-linked holdings | Money market and gold provide some ballast |
| Stagflation | Gold, Infrastructure | Growth equities, REITs | Better than a 60/40, still difficult — the combined ~20% in gold and infrastructure is the main ballast |
What the table is there to show is that the repression thesis is treated as a base case, not the only future. The mitigant is that 60% public equity participates meaningfully in most non-repression outcomes, so the portfolio isn't a one-way bet — only the gold and real-asset layer is genuinely regime-concentrated.
The portfolio describes its own reasoning as Black-Litterman-style. That claim doesn't fully survive examination, and it's worth saying so rather than letting it stand.
Black-Litterman starts from market-cap equilibrium weights and blends in views, each carried at an explicit confidence level. A standard global multi-asset equilibrium prior holds something like 35–40% fixed income. Getting from there to zero requires asserting the repression view at near-certainty — because any confidence below 100% mathematically nets out to a probability-weighted blend that still retains some bond weight.
Near-certainty is a much stronger claim than the roughly 75–80% confidence that's honestly defensible for a multi-decade macro view. Run with genuine confidence levels, the posterior would land somewhere around 12–15% fixed income, not 0%.
The honest conclusion: this portfolio uses Black-Litterman reasoning — an equilibrium starting point plus explicit named views — but its zeros are maximum-conviction overrides, not confidence-weighted model output. They rest on separate, specific arguments, and the Black-Litterman language properly applies to the relative weights of everything else.
Duration isn't the same thing as repression. Short and intermediate bonds bought at reasonable yields can still deliver positive real returns under mild repression. The exclusion argument really applies to long-duration nominal bonds specifically, not to fixed income as a uniform category — the 0% is blunter than the reasoning behind it.
Factor definitions carry unverified baggage. Quality, Value and Momentum as implemented by a specific index provider may carry sector or regional biases distinct from the academic literature used to justify the tilt. That's a real gap, not a rhetorical concession — it warrants checking the funds' actual sector breakdowns against a plain cap-weighted index before claiming precision about factor exposure.
Rebalancing is annual and drift-based rather than mechanical percentage bands — trim what outperformed, add to what lagged. That enforces buying low and selling high without reintroducing the discretionary market timing the framework rejects.
The metric that matters most, ahead of capital weight — it surfaces the concentration a weights table hides.
Rather than a static matrix, because regime shifts are what break a diversification assumption. The 2022 bond–equity flip is the obvious example.
To compare risk-adjusted contribution across parts of the portfolio that aren't otherwise comparable.
Each satellite carries one, tied to the thesis that justified it — a leverage threshold, a regulatory change, or a reversal in the structural driver.
This is a live framework rather than a finished analysis. What it's taught me is mostly about the difference between a view and a position: a macro thesis is easy to hold in conversation and much harder to hold once it has forced you to zero out an entire asset class and then watch a scenario unfold where that was the wrong call.
The parts I'd defend hardest aren't the weights — they're the places where the reasoning is written down honestly enough that someone could attack it properly.
Note: this write-up covers allocation logic and sleeve weights. It deliberately does not disclose position sizes, absolute amounts, individual holdings, or realized returns.