Model portfolios · Simulated data · illustration only

Three Allocations, One Regime Model

Diversified, conservative allocations whose weights are set by the same overlays you see on the chart. Fully published, rebalanced monthly, sized to be held through a bad year. A framework to study and adapt — not a recommendation.

The framework needs one input: which regime you are in. Hand that research to the analyst. It splits the correlation on all three axes, says which bucket the last session sits in and since when, and shows what the split changes — the regime read →

Current regime: Disinflation / easing · liquidity easing since 14 Jun 2026 · next review 2026-10-01
Disinflation / easing · CPI trend falling, Fed Funds falling, real yields ≤ 1.5 %, escalation < 70 · weights at the 2026-09 rebalance

Disinflation / easing current

Inflation trending down, policy rate falling, liquidity improving. Risk assets carry the most weight the framework ever allows; duration is rewarded; gold stays as the hedge against the regime being wrong.

SleeveTargetCurrentDriftRule that changes it
Global equities40 %41.6 %+1.6−10 if CPI trend turns up 2 prints running
Government bonds 7–10y25 %24.1 %−0.9−15 if Fed Funds direction flips to hiking
Gold15 %15.3 %+0.3+5 if escalation ≥ 70 for 4 weeks
Broad commodities5 %4.8 %−0.2+20 on reflation regime
Digital assets (BTC / ETH)5 %5.4 %+0.40 on risk-off; capped at 5 always
Cash / T-bills10 %8.8 %−1.2absorbs every cut above

Rebalance band ±3 pts · nothing is outside the band today, so the September rebalance made no trades. Drift is measured against the target since 2026-08-01.

Reflation / hiking · CPI trend rising, Fed Funds rising or on hold at the peak, commodities leading · last active 2021-10 → 2023-11

Reflation / hiking

Inflation trending up, policy tightening, real assets leading. Duration is cut hard, commodities carry the tilt, cash is the largest it ever is because the framework expects both stocks and bonds to struggle together.

SleeveTargetvs easingRule that changes it
Global equities30 %−10−10 more if 2s10s inverts
Government bonds 7–10y10 %−15+15 when Fed Funds direction flips to cutting
Gold15 %0+5 if real yields fall below 0
Broad commodities25 %+20−10 if COT crude z > +2 (crowded)
Digital assets (BTC / ETH)5 %00 on risk-off
Cash / T-bills15 %+5absorbs every cut above

Last active 2021-10 → 2023-11 (25 months). Entered when GDELT INFLATION tone crossed −2σ and CPI printed above 5 % for the second month; exited when the Fed Funds direction flag turned to "hold at peak, cuts priced".

Contraction / risk-off · ISM < 48 two months, payroll trend negative or escalation ≥ 85, credit spreads widening · last active 2020-03 → 2020-06

Contraction / risk-off

Growth rolling over or a geopolitical shock at the top of the scale. Duration and gold carry the portfolio; equities are the smallest they ever are; the digital-asset sleeve goes to zero because it has not behaved like a hedge in any risk-off window in the record.

SleeveTargetvs easingRule that changes it
Global equities20 %−20+20 back when ISM > 50 and spreads narrow 2 months
Government bonds 7–10y35 %+10−10 if CPI trend rises during the contraction
Gold20 %+5held
Broad commodities5 %0held
Digital assets (BTC / ETH)0 %−5restored at 5 on any other regime
Cash / T-bills20 %+10absorbs every cut above

Last active 2020-03 → 2020-06 (4 months). Entered on the ISM and payroll flags together with an escalation-scale spike in the global tone index; exited when both growth flags cleared.

See the weights in the terminal →
Method

How the Regime Model Sets Weights

Four flags from the same overlays on your chart. The combination picks one of three allocations; the rules in the tables above adjust it at the edges. No forecast anywhere in the chain.

Flag 1 · FRED

Inflation trend

Three-month change in CPI YoY, release-aligned. Up = reflation pressure, down = disinflation. Two consecutive prints are needed to flip.

Flag 2 · FRED

Policy direction

Fed Funds: hiking, hold-at-peak, cutting. Read from the effective rate and the FOMC calendar, never from expectations.

Flag 3 · FRED

Growth

ISM manufacturing below 48 for two months or a negative three-month payroll trend, together with widening credit spreads, flags contraction.

Flag 4 · ACLED · GDELT

Geopolitical shock

Escalation index ≥ 85 in any region, or ≥ 70 for four weeks, adds a watch flag: gold +5 in any regime; ≥ 85 forces risk-off.

Regime history 2019–2026 · shaded bands: risk-off (red) · reflation (gold) · easing (green) · line: CPI YoY, release-aligned

Why three, why so conservative

  • Three regimes, not thirty. More states means more flips, more trades and more ways to be wrong. Three has been enough to move the big weights when it mattered and stay still the rest of the time.
  • No sleeve above 40 %, no leverage, cash never below 10 %. A framework has to survive the year in which it is wrong. These caps are the price of that.
  • Monthly, banded. Rebalance on the first business day only if a sleeve is more than 3 points off target. Most months nothing trades.
  • Every rule is a recipe. The four flags are overlay recipes shipped with the terminal, so you can see the flag on your own chart the day it changes — and change the rule if you disagree.
The analyst · regime panel

Hand the Regime Read to It

The framework above needs one input from you: which regime you are in. That is research, and it is exactly what the analyst does. It splits the correlation on each of the three axes, tells you which regime the last session sits in and since when, and — the part that matters — how much the split changes the answer. It reads the regime; it does not set a weight and it does not size anything.

AxisLabelled fromr across all 10 yearsr inside regimesSpread the single number hidesNow
LiquidityFed Funds, 63-session change−0.36−0.47 … −0.250.22Easing since 14 Jun 2026
InflationCPI YoY, 126-session change+0.47+0.29 … +0.610.32Steady since 25 Jun 2026
Geo-riskACLED escalation, 63-session mean+0.42+0.24 … +0.560.32Elevated since 24 Aug 2026

XAUUSD against the series each axis is about, ten-year sample. Ask the analyst is the split real? and it recomputes the correlation inside every regime bucket; buckets under 60 sessions are dimmed and flagged as thin rather than quoted.

What the geo-risk split actually says

Geo-risk regimeSessionsXAUUSD return / yrVolr vs escalation
Calm602+13.8 %13.2 %+0.24
Normal1,238+13.2 %14.8 %+0.42
Elevated680−5.9 %17.0 %+0.56

The escalation index is at 88 and the last session sits in the Elevated bucket — the one where gold's ten-year annualised return is negative and its volatility is highest. That is a description of 680 past sessions, not a forecast for the next one.

What it will and will not do here

  • It will label the regime, say since when, and show what the label does to the correlation — on all three axes, with the sample size behind each bucket.
  • It will arm a condition on the reading that flips the label, backtest it over 2,520 sessions and tell you how often it crossed — then watch it and log what it finds, including the checks where nothing crossed.
  • It will not propose a weight, a trade size or a return. Asked, it answers plainly: "RegimeLens is a research terminal, not an execution or advice tool." The allocations above are a published framework you read and adapt; the analyst is the research that tells you which column you are in.
  • It will not overstate the split. Where the gap between regimes is under 0.10 it says so — "on this axis the split barely matters, which is itself worth knowing."

Questions

Is this investment advice?
No. The model portfolios are an educational framework that shows how a rules-based, diversified allocation can be tied to observable macro data. They are not tailored to you, and they are not a recommendation to buy or sell anything. Talk to a licensed adviser about your own situation.
Where are the returns?
We publish weights, rules, regime history and drift — the things you can verify. We do not lead with performance figures because they invite exactly the wrong comparison. Export the weights (Pro) and test them in your own tools with your own assumptions about costs and instruments.
Why no leverage and why is cash never below 10 %?
Because the framework is built to be held through the year in which it is wrong. Leverage turns a bad regime call into a permanent loss; cash and caps turn it into a bad year. We would rather compound slowly for a decade.
How often do the weights actually change?
Regimes flipped five times between 2019 and 2026. Within a regime, the monthly rebalance trades only when a sleeve drifts more than 3 points; in the current regime that has been three of fifteen months.
Can I change the rules?
Yes. The four flags are overlay recipes in the terminal; edit the thresholds, add a flag, or replace a sleeve. The drift table recomputes against your version.

Model portfolios are illustrative allocations for educational purposes. Nothing on this page is investment advice or a recommendation. Diversification does not guarantee against loss; past regime behaviour does not predict future behaviour. RegimeLens is analysis software and does not execute trades.