Stress · Live · 5 scenarios
Historical Stress Tests
Current weights replayed through 5 historical crises — where would each group break?
Insurance / wealth IC committees run this internally — published openly here.
| Scenario | Period | Market drawdown | 60/40 | Base | Bull | Bear |
|---|---|---|---|---|---|---|
| 2008 GFC | 2008-09-15 → 2009-03-09 | S&P −56% · 信用利差 +500bp · USD 高位 | −24% | −12.4% | −18.7% | −4.2% |
| 30% duration + 5% gold cushioned half the drawdown; A-shares offered limited diversification (-65% in 2008) | ||||||
| 2020 COVID liquidity crisis | 2020-02-19 → 2020-03-23 | S&P −34% (32 天) · VIX 峰值 82 · 流动性枯竭 | −21% | −15.8% | −21.3% | −7.1% |
| all correlations spiked to 1; even gold sold off briefly; cash + short duration were the only defense | ||||||
| 2015 CNY devaluation + A-share crash | 2015-06-12 → 2016-01-28 | CSI 300 −45% · CNY −5% · 黄金 +6% | −6% (美国侧) | −9.6% | −14.2% | −1.8% |
| 30% CN equity took direct hit; gold + US Treasury saved the book; US equity barely moved | ||||||
| 2022 inflation shock (stock-bond double dump) | 2022-01-03 → 2022-10-12 | S&P −25% · UST 20Y+ −31% · 60/40 −21% | −21% | −18.2% | −22.5% | −14.8% |
| the death scenario for risk parity — stocks + bonds correlated, gold weak; even bear couldn't escape. Commodities the only hedge | ||||||
| 1998 LTCM + Russia default | 1998-08-17 → 1998-10-08 | S&P −19% · EMG 债 −40% · 信用利差 +400bp | −9% | −6.3% | −9.8% | −1.2% |
| EM + credit cratered, but US equity bounced fast; long duration rallied; base loss contained | ||||||
Worked example: 2008 GFC under current weights
base
Base group simulated: −12.4% (vs 60/40 −24% — long duration 30% + 5% gold cushioned half the drawdown)
bull
Bull group simulated: −18.7%
bear
Bear group simulated: −4.2% (60% cash + duration + gold saved it)
Example uses 2026-06 weights × 2008 asset returns; framework validation only, not live performance.
methodology
Return simulation method (v0.1, to be replaced by real interval returns once stress script lands): 1. Take current (2026-06-01-inception) weights 2. For each ETF, estimate scenario-period total return in local currency (dividends included) 3. Offshore assets keep local currency (no USD/CNY adjustment) — consistent with v1 NAV methodology 4. Weighted sum: r_group = Σ (w_i × r_i) 5. Excludes commissions / slippage / tax
Full methodology in methodology.md §8 Stress Test Methodology (draft).