| Scenario | Probability | Key Driver | Market Impact |
|---|---|---|---|
| Base: ELEVATED continues | 55% | VIX 16–22, gradual easing narrative holds | Modest spread tightening, curve steepening. Carry strategies perform. Equity grinds higher with rotation. |
| Upside: Return to NORMAL | 25% | VIX < 16, strong data, dovish Fed | Risk rally, spread compression (IG → 90bp, HY → 340bp). Equities +3–5% from here. |
| Downside: Shift to STRESS | 18% | VIX > 25, geopolitical shock, hot CPI | Spread widening (HY → 500bp+), flight to quality. Duration rallies, credit sells off. Risk-off across EM. |
| Tail: CRISIS | 2% | VIX > 35, systemic event, liquidity shock | Liquidity seizure. HY > 800bp. Treasuries bid massively. Equity -15%+ drawdown. Correlation spike. |
| Asset | Base (55%) | Upside (25%) | Downside (18%) | Tail (2%) |
|---|---|---|---|---|
| 10Y Treasury | 4.15 – 4.45% | 4.40 – 4.60% | 3.80 – 4.10% | 3.20 – 3.60% |
| IG OAS | 95 – 110bp | 80 – 95bp | 130 – 160bp | 200 – 300bp |
| HY OAS | 350 – 400bp | 300 – 350bp | 500 – 650bp | 800 – 1,200bp |
| S&P 500 | +5 to +8% | +10 to +15% | -5 to -15% | -20 to -35% |
| EUR/USD | 1.06 – 1.10 | 1.10 – 1.14 | 1.02 – 1.06 | 0.95 – 1.02 |
| Gold ($/oz) | $2,900 – $3,100 | $2,700 – $2,900 | $3,100 – $3,400 | $3,400 – $4,000 |
| VIX | 15 – 20 | 12 – 16 | 22 – 35 | 35 – 80 |
| Sector | Rating | Momentum | Valuation | Catalyst | Trade Idea |
|---|---|---|---|---|---|
| AI / Semiconductors | OVERWEIGHT | Strong ▲ | Stretched | Earnings, capex cycle | Long NVDA/AVGO, hedge via QQQ puts |
| Energy | UW → NEUTRAL | Weak ▼ | Cheap | OPEC+, demand recovery | Selective: quality integrated names only |
| Healthcare | OVERWEIGHT | Neutral – | Attractive | Drug pricing, M&A | Defensives with pipeline optionality |
| Financials | NEUTRAL | Positive ▲ | Fair | NII trajectory, credit quality | G-SIBs over regionals; short CRE-exposed |
| Real Estate | UNDERWEIGHT | Weak ▼ | Distressed | Rate cuts needed for relief | Avoid office; selective industrial/logistics |
| Utilities | OVERWEIGHT | Positive ▲ | Attractive | Data centre power demand | Regulated + renewables exposure |
| Consumer Discretionary | UNDERWEIGHT | Neutral – | Fair | Consumer sentiment, rates | Avoid leverage; favour luxury + experience |
| Metric | 1-Month | 3-Month | 6-Month | 12-Month |
|---|---|---|---|---|
| Regime direction | 78% | 65% | 52% | 41% |
| Regime timing (±2 wks) | 62% | 48% | 35% | 28% |
| VIX direction | 71% | 58% | 49% | 43% |
| Spread direction | 74% | 61% | 51% | 39% |
MSE regime classification is most accurate at 1-month horizons (78% directional accuracy) and degrades at longer horizons as it should — markets are not predictable beyond short windows. This is a feature, not a bug.
The model excels at identifying regime persistence — once in STRESS, the probability of remaining in STRESS is well-calibrated. It struggles most with regime transitions, specifically the exact timing of shifts between states.
Notable successes: COVID March 2020 was detected within 3 trading days. The 2023 SVB crisis triggered STRESS classification within 24 hours. Both were driven by VIX spike + credit spread dislocation signals.
Notable failures: The 2022 rate shock took 2 weeks to fully classify as STRESS because the HMM was trained on vol-driven regimes, not rate-driven regime changes. The model also generated a false STRESS signal in June 2024 that reverted within 4 days.
Areas for improvement:
(1) Rate regime vs credit regime differentiation
(2) Transition speed detection — distinguishing fast vs slow regime shifts
(3) Incorporating forward-looking indicators (PMI, earnings revisions, fund flows)
(4) Sector-level regime classification (tech vs energy can diverge)
Accuracy metrics computed via walk-forward backtest with 252-day training window, no look-ahead bias. See WP-RQ-015 for full methodology.
| Risk | Probability | Severity | Monitor | Trigger Level |
|---|---|---|---|---|
| US recession | 20% | HIGH | ISM PMI, initial claims, consumer confidence | PMI < 48, claims > 300K sustained |
| China hard landing | 10% | HIGH | Caixin PMI, property sales, credit growth | PMI < 47, property sales −30% YoY |
| Geopolitical escalation | 15% | VERY HIGH | VIX, gold, oil, defence stocks | VIX > 30, oil > $90, gold > $3,300 |
| Credit cycle turn | 25% | MEDIUM | HY OAS, default rate, leveraged loan distress | HY OAS > 500bp, trailing defaults > 3% |
| Inflation re-acceleration | 15% | MEDIUM | Core CPI, Core PCE, wage growth, TIPS breakevens | Core PCE > 3.5%, breakevens > 2.8% |
| Japan policy shock | 10% | HIGH | USD/JPY, JGB 10Y, BoJ communications | JPY breaks 160, JGB 10Y > 2%, carry unwind |