Scenarios · one macro force at a time
If market selloff hit — s&p 500 falls 10% — the engine estimates the book would lose about −7.9% (−$14,848), led by KGC. The friendliest shock, market rally, would add about +4.0%. Each scenario moves one force at a time — a real crash moves several at once.
Worst · Market selloff
−7.9%
Best · Market rally
+4.0%
Scenarios run
7
first-order, single-factor · each holding’s How strongly a holding moves when a macro force moves — measured from how the two have actually traded together. measured over 120 trading days · 10y real yield 2.41% (DFII10, 2026-07-30)
For crashes where several forces move together, the historical stress library lives on Risk.
Risk →Worst first. Each row is one force snapping while everything else holds still.
| Scenario | Move | Est. book | USD |
|---|---|---|---|
| Market selloff | S&P 500 falls 10% | −7.9% | −$14,848 |
| Dollar strengthens | US Dollar +5% | −6.1% | −$11,441 |
| Real yield +25bps | 10-year real yield rises 25bps | −3.3% | −$6,205 |
| Oil spikes | Crude oil +20% | −2.2% | −$4,045 |
| Gold rally | Gold +10% | +3.1% | +$5,877 |
| Real yield −25bps | 10-year real yield falls 25bps | +3.3% | +$6,205 |
| Market rally | S&P 500 rises 5% | +4.0% | +$7,426 |
S&P 500 falls 10% · est. −7.9% (−$14,848)
Lead drivers
S&P 500 rises 5% · est. +4.0% (+$7,426)
Lead drivers
Move one factor and watch the estimated book impact respond.
Market shock
-10%Estimated book impact
−7.9%
−$14,833
Led by KGC −2.1%
First-order estimate: book move = Σ weight × beta × shock, one factor at a time. An estimate of the systematic response, not a forecast.
Method: each holding's response is beta to the shocked factor times the shock, where beta = correlation × (holding volatility ÷ factor volatility), measured over the last 120 trading days. The book impact is the weighted sum across positions; cash is unaffected. Limits: first-order and single-factor — it ignores cross-factor moves, second-order effects, and the fact that in a real shock several forces move at once. Betas are regime-dependent and the relationships can break in a true tail. This is an estimate of the systematic response, not a forecast.
Real yield: a yield is a level, not a price, so the rate rows use a separate measure — each holding's daily return regressed on the daily change in the 10-year real yield (DFII10, from FRED), giving a beta in return per 100bps. Measured over 120 overlapping days. Kept out of the correlation matrix because the two are not on the same scale.
