Scenario Testing takes your current open book (every position, every size) and replays it through curated historical crises. The 2008 financial crisis. The 2010 Flash Crash. The 2015 SNB shock. COVID March 2020. The 2022 yen intervention. You see the day-by-day P&L impact, the worst drawdown moment, the recovery (or not). Stop wondering. Start knowing.
If you started trading any time after 2020, you haven't really seen markets break. Everything you've experienced is either calm, mildly volatile, or a single-direction trend. The day correlations go to one, liquidity vanishes, and your three "diversified" positions all gap together. That day is invisible in your live experience.
Risk management designed only against the markets you've seen isn't risk management. It's surviving on the assumption that the future looks like the recent past.
Scenario Testing solves this by giving you instant access to the moments markets broke. Historical crises, curated for tradability, replayed against your current book at the level of detail that matters. Not a vague "what if there's a crash." A specific, dated, dollar-quantified "this is what your current portfolio loses on the third day of the COVID crash, by position, by hour." Then you adjust before, not after.
Every scenario is fully reconstructed with tick-level data where available, daily resolution elsewhere. Each comes with context: what happened, why it happened, which instruments moved most, and how long until recovery (if at all). You can run any scenario on your current book in 3 seconds.
Lehman Brothers fails. Credit markets freeze. Cross-asset correlations approach 1.0. Most "diversified" portfolios behave as one trade.
Largest intraday move in S&P history (at the time). 9% drop and full recovery in 36 minutes. Test your liquidity assumptions.
The Swiss National Bank abandons its EURCHF 1.20 floor with no warning. EURCHF moves 30% in minutes. Many brokers bankrupt overnight.
Fastest 30%+ S&P drop in history. Volatility regime shift in days. Oil futures go negative. Everything that "shouldn't correlate" correlated.
First Japan FX intervention since 1998. USDJPY drops 5.4 yen in hours. Carry trades unwind violently. Cross-yen pairs see record vol.
Japanese rate hike triggers global carry-trade liquidation. Yen crosses gap 5–8%. SPX drops, BTC drops, gold drops. Every "safe haven" sells off together.
Take your current open positions exactly as they are. Apply any historical scenario. See the day-by-day, hour-by-hour P&L impact. No assumptions, just substitution.
Total scenario P&L is decomposed by position. Find out which trade is doing most of the damage. Often the surprise: the "safe" hedge that wasn't.
Across the scenario, when was the deepest drawdown? Down to the minute. See whether your account would have margin-called, whether stops would have been jumped, whether you'd have survived.
Define your own shock: "USD spikes 8%, gold drops 12%, BTC drops 30%, correlations all go to 0.9." Run it. See what you'd lose. Useful for what-if analyses no historical replay covers.
Side-by-side comparison: COVID vs. SNB vs. 2008. Same book, different crises. Discover which kind of break is the one to fear in your specific configuration.
at.backtest(..., scenarios=["covid_2020"]). Pre-deploy gate: any strategy with intolerable scenario losses is auto-flagged before going live.
Defaults to your current live positions. You can also load a hypothetical book, a saved configuration, or a strategy's typical book from Strategy Lab.
Choose from 11 curated historical events or define a custom shock. Adjust parameters: scale severity, change duration, modify correlation behavior. Defaults reflect history exactly.
Day-by-day P&L curve, worst-moment metrics, per-position attribution, margin / stop-out analysis. Share the report. Adjust the book. Re-run.
Risk Exposure flags concentration. Scenario Testing tells you what that concentration loses in 2008. They combine as the Risk Pack bundle.
See Risk Exposure →VaR gives you "the worst 1-in-20 day." Scenario Testing gives you "the worst 1-in-1000 day." Two different lenses on the same exposure.
See VaR Analysis →Strategy Lab's Scenario Lab uses this engine. Every strategy gets graded against historical crises before live deploy. Strategies that blow up in 2008 stay in research.
See Strategy Lab →Scenario Testing pays for itself on the single trade you don't take because the scenario report told you what would happen if markets broke.
11 curated historical crises. Custom scenario builder. Tick-level replay where data exists. Per-position attribution. Worst-moment finder. Strategy Lab integration. The same engine the Crisis Replay system in Strategy Lab runs on.
14 days free. No card. Full app from day one.