Case Study — Structural Limitation
Japan: the structural limitation we disclosed
Japan is classified live under the v2.1 framework, and the structural limitation we disclosed when its backtest numbers were weakest remains instructive. Understanding it is more valuable than any single performance number.
The result
Under the v2.1 production framework, Japan’s backtested portfolio Sharpe is 0.793 with theory-first ETF baskets — the softest of the six regions, reflecting basket behaviour under Japan’s rate history rather than classifier failure. Earlier “significance gate” claims for this region used a permutation method we have since withdrawn; see the Performance page for where statistical validation stands.
But the historical structural limitation that characterized Japan's backtested performance remains worth understanding. Before rate normalisation, example baskets applied to the Japanese market produced a CAGR of −0.2% versus +3.8% for buy-and-hold. The signal worked; the defensive carry didn't. We published this case study when Japan was unvalidated, and we preserve it now because the analysis reveals what regime classification is — and what it is not.
Two separate problems
1. The baskets are structurally wrong for Japan
RegimeR's example defensive baskets assume bonds rally during equity stress. In the US, Treasuries gained 20–30% during the GFC as capital fled to safety. In Japan, JGBs yielded 0–1% for 25 years. There was no room to rally. The defensive “switch to bonds” had zero carry and near-zero upside.
Cash yields were negative under NIRP. Holding cash cost money. Gold provided some offset but couldn't compensate for the combined drag of zero-yield bonds and negative-yield cash. In every other covered region, defensive assets generate positive carry that funds the opportunity cost of being out of equities. In Japan, they don't.
2. A missing signal: the yen
Japanese equity returns are tightly coupled to yen weakness. When the yen depreciates, exporters' earnings rise and the Nikkei rallies. When the yen strengthens, the opposite. This FX-equity dynamic is the dominant driver of Japanese equity returns — more so than credit stress or growth momentum.
RegimeR's classifier does not include FX positioning. During Abenomics (2013), the dominant market dynamic was yen weakness driving equity returns — the Nikkei surged +26% on currency depreciation. This is a currency-driven pattern the macro classifier does not capture as a distinct regime. The classification reflected mixed macro conditions, which was accurate, but missed the dominant driver of Japanese equity performance in that period.
What the signal did detect
During genuine crises, the regime classification was directionally correct:
| Episode | Nikkei | Example basket | Difference |
|---|---|---|---|
| GFC (2007–2009) | −47.4% | −18.6% | +28.8pp |
| COVID (2020) | −4.1% | +3.6% | +7.7pp |
| Abenomics (2013) | +26.0% | −1.3% | −27.3pp |
The GFC and COVID results show the signal detecting genuine stress correctly. Even with suboptimal baskets, the backtested defensive allocations preserved 29 and 8 percentage points respectively. The Abenomics result shows the opposite problem: the signal read mixed macro conditions correctly, but the Japanese equity market rallied on currency dynamics the classifier does not capture as a distinct regime.
Why this validates the model
Japan proves that the regime signal and the defensive response are genuinely separate. The signal detected real regime shifts. The example baskets failed because Japan's zero-rate environment eliminates the defensive carry that makes basket switching profitable in other markets.
A Japanese institutional investor receiving “Contraction regime detected” would not switch to JGBs at 0.5% yield. They would deploy Japan-appropriate hedges: yen longs, Nikkei put options, or reduced cross-currency exposure. The regime signal has value. The example basket does not. That distinction is the core of what RegimeR provides.
What's changing
Japan ended negative interest rates in March 2024. The BOJ policy rate is now 0.5% — the highest since 2008. JGB 10Y yields have risen above 1% for the first time in over a decade. Japan is experiencing inflation (CPI above 2%) for the first time in a generation.
Japan's rate environment has normalised. The BOJ policy rate is now 0.5% — the highest since 2008. JGB 10Y yields have risen above 1%. Under the v2.1 framework with theory-first baskets that account for the rate normalisation, Japan’s backtested results improved materially. The structural limitation was real but time-bounded.
Could this happen in other regions?
The six live regions (US, AU, DE, GB, JP, CA) all maintain functioning sovereign bond markets with positive nominal yields. None have experienced the sustained zero-rate environment that characterises Japan's structural limitation. Germany operated under ECB negative rates from 2019–2022, but bund yields still rallied during Eurozone stress events (flight to Germany from periphery) — a dynamic that JGBs do not benefit from. If a covered region were to adopt sustained ZIRP, the example baskets would need to be redesigned — but the regime signal would remain applicable.