Canada: The Oil-Exporter Problem
Energy dynamics work differently for commodity exporters. Rising oil prices are inflationary stress for importers but revenue growth for exporters. The system discovered this distinction and adapted.
The problem
Most macro regime classifiers treat energy prices the same way everywhere: rising energy costs signal inflationary stress. That assumption is correct for importers — the US, Germany, the UK, Japan. It is wrong for exporters.
Canada is a major oil exporter. When crude prices rise, Canadian export revenues increase, the currency strengthens, and equities in the resource sector rally. The stress signal for importers is a growth signal for exporters. Applying the same energy logic to both produces systematic misclassification.
When oil prices collapse — as they did in 2014–2016 — the inverse applies. Falling oil is a tailwind for importers but a direct hit to Canada’s terms of trade, fiscal position, and equity market.
Three episodes, three different dynamics
| Episode | Period | 60/40 | RegimeR | Net |
|---|---|---|---|---|
| GFC oil crash | Jul–Dec 2008 | -22.5% | +1.4% | +23.9pp |
| Oil rout | 2014–2016 | -2.6% | -4.0% | -1.4pp |
| COVID oil crash | Jan–Apr 2020 | -5.4% | +3.0% | +8.4pp |
The 2014–2016 oil rout: the one we got wrong
When oil prices fell 70% between mid-2014 and early 2016, Canada’s equity market declined but credit markets remained calm and the broader economy avoided recession. The system classified this as a transition rather than a contraction — and the backtested defensive allocation underperformed 60/40 by 1.4 percentage points.
This is the only episode across all 16 backtested events where RegimeR underperformed the benchmark. We include it because honest reporting of failures is more credible than a perfect record.
What the system discovered
In backtested validation, the optimisation process independently identified that energy signal polarity needed to be inverted for commodity exporters. Rising oil prices, which trigger defensive classification in importing economies, were correctly identified as supportive for Canadian markets.
This was not a design decision made in advance — it was a result discovered systematically through the validation process. A global model using uniform energy assumptions would have misclassified Canadian market conditions in every oil-driven episode.
Validation
1.387
Sharpe ratio (backtested)
p < 0.0001
Permutation test (30,000 simulations)
Canada produced the strongest statistical result of any validated region, with a z-score above +6.6 — meaning the regime signal’s outperformance is over six standard deviations above random. All figures are backtested.