Scenario analysis for equity portfolios
See how your book behaves before the scenario happens.
Headwinds turns a macro narrative into factor shocks, prices them against every holding, and shows the P&L distribution with the reason behind each line.
Or open the sample workspace. US-listed equities and ETFs, one to three month horizon.
How it works
01
Upload the book
A CSV of tickers with shares or weights. Names, sectors and prices are resolved against the nightly universe, and anything we cannot price is flagged, not dropped.
02
Set the scenario
Pick from the library or write it in a sentence. The narrative becomes a short vector of factor shocks with its assumptions spelled out. Edit any number before you run.
03
Read the result
A P&L distribution with P5, P50 and P95, the winners and losers, and one line per holding naming the factor that drives it. Then replay the same book through history.
A worked example
Illustrative numbers from the sample book
“The Fed cuts 100bp and oil goes to $120.”
Two factors are set. The rest are implied from how the factors have moved together.
| Factor | Proxy | Shock | Mode |
|---|---|---|---|
| Rates | IEF | +3.0% | set |
| Oil | USO | +85% | set |
| Broad equity | SPY | −2.4% | implied |
| Credit | HYG − IEF | −0.6% | implied |
| Inflation expectations | TIP − IEF | +1.9% | implied |
| Dollar | UUP | −1.1% | implied |
P5
−7.4%
Expected
−1.8%
P95
+3.6%
Helped most
| XOM | Exxon Mobil | +6.1% | Oil |
| TLT | 20+ Year Treasury | +5.2% | Rates |
| FCX | Freeport-McMoRan | +4.8% | Oil |
Hurt most
| NKE | Nike | −7.9% | Oil |
| HD | Home Depot | −6.4% | Broad equity |
| NVDA | NVIDIA | −5.8% | Broad equity |
Methodology
A factor model you can explain in a meeting.
Nine market-observable factors. Equity, size, rates, curve, credit, inflation expectations, oil, the dollar and equity vol, each proxied by a liquid instrument with a daily history. No revised macro series, no judgment calls hidden in the data.
Shock completion. A scenario sets some factors. The others take their conditional mean under the fitted covariance, so a rates-only scenario still moves credit and equities the way history says they co-move. The completed vector is shown, every number editable.
Holding by holding. Each name carries betas to the nine factors, shrunk toward its sector, plus a residual risk of its own. The P&L distribution combines both. Attribution is weight times beta times shock, one number per factor, so every line on the results page has a cause.
What it is not. Not a forecast of the scenario itself, not a multi-period path, and not investment advice. The model is validated by replaying books through five historical windows:
- COVID crashFeb to Mar 2020
- Hiking cycleJan to Oct 2022
- Q4 2018 selloffOct to Dec 2018
- SVBMar 2023
- Tariff shockApr 2025