Case Studies

Selected stress events examined in detail. Full 13-episode results on the Performance page.

2007–2009+34.8pp average preservation across 4 regions

The Global Financial Crisis

The defining credit cascade — subprime, bank failures, 50%+ equity drawdowns. In backtested analysis, RegimeR detected the regime shift and would have preserved an average of 34.8 percentage points of value relative to 60/40.

2022Our hardest test — honestly assessed

The 2022 Rate Shock

Bonds and equities fell simultaneously for the first time since the 1970s. RegimeR outperformed 60/40 in every region but still lost money in the US.

September 2022+13.1pp preservation — discovered systematically

The UK Gilt Crisis

The Truss/Kwarteng mini-budget triggered a 42% gilt collapse. In backtested validation, RegimeR's quantitative optimisation independently set UK gilt allocation to zero in defensive regimes — a result discovered systematically, not chosen in advance.

2008–2020Strongest z-score of any region — and our only underperformance

Canada: The Oil-Exporter Problem

Energy dynamics work differently for commodity exporters. The system discovered the polarity inversion and adapted. Strongest validation result (z=+6.66) across 3 episodes, including the 2014 oil rout — the only episode where RegimeR underperformed 60/40.

2008–2022Commodity amplifier, commodity buffer, rate shock

Australia: Three Crises, Three Mechanisms

Each stress episode hit Australia through a different channel. The GFC amplified losses through commodity exports. COVID was cushioned by commodity recovery. The 2022 rate shock hit bonds and equities simultaneously. Region-specific classification adapted to each.

1998–2021Signal works, example baskets don't — published transparently

Japan: A Structural Limitation

25 years of zero interest rates eliminated the defensive carry that makes basket switching profitable. The regime signal detected real stress, but standard baskets couldn't capitalise. Japan proves the signal and the response are genuinely separate.