How does a historical scenario replay work?

The most concrete stress test available is also the simplest: take a crisis that really happened and run today's portfolio through it. No distributional assumptions, no probability arguments — just the actual moves, applied to the actual positions you hold now.

The method in four steps

1. Choose the episode and its exact window. Not "2008" — 15 September to 31 December 2008. Precision matters, because the answer changes with the dates.
2. Collect the actual moves of every risk factor your portfolio touches over that window: equity indices, credit spreads, government yields, currencies, commodities, volatility.
3. Apply those moves to today's holdings. Your current portfolio, the historical shocks.
4. Read the total. That is the scenario loss.

The output is a sentence of the form: "If the fourth quarter of 2008 repeated on today's book, we would lose X."

A standard scenario library

Most risk functions maintain a set of replays. Common episodes and their headline equity moves (illustrative, rounded):

• Black Monday, 19 October 1987 — Dow −22.6% in one session. Tests single-day gap risk and any position that assumes you can trade on the way down.
• Global Financial Crisis, 2007–2009 — the S&P 500 fell roughly 57% from its October 2007 peak to its March 2009 trough. Tests deep, prolonged drawdown and funding stress.
• COVID crash, 19 Feb – 23 Mar 2020 — the S&P 500 fell about 34% in 33 days. Tests speed: how much can be repositioned when the whole move takes five weeks.
• 2022 rates shock — the S&P 500 fell about 18% on a total-return basis while the Bloomberg US Aggregate bond index fell about 13%, its worst calendar year on record. Tests portfolios built on the assumption that bonds cushion equity losses.

Note the last one especially. It is in the library precisely because it broke the pattern the other three shared.

A worked replay

Take a simple portfolio: 60% broad equity index, 40% aggregate bonds, total value $1,000,000.

Apply the 2022 moves:

• Equities: 60% × −18% = −$108,000
• Bonds: 40% × −13% = −$52,000
• Total: −$160,000, or −16.0%

Now apply 2008 to the same portfolio, using the calendar year:

• Equities: 60% × −37% = −$222,000
• Bonds: aggregate bonds rose about 5% in 2008 → 40% × +5% = +$20,000
• Total: −$202,000, or −20.2%

Two crises, two entirely different mechanisms. In 2008 the bond allocation cushioned the fall. In 2022 it added to it. The same portfolio has a different risk profile depending on which historical regime you replay — and that variation is the most valuable output of the whole exercise.

The limits, stated plainly

The next crisis will not be a rerun. Every historical scenario is a sample of one, drawn from a specific structure of markets, participants and policy that no longer exists in the same form.

Instruments change. Products that exist today did not exist in 1987; mapping historical factor moves onto them requires assumptions.

Policy responses differ. Central banks responded to 2020 within weeks; the 1930s response was slower and different in kind. A replay implicitly assumes the response you observed.

The window choice is a judgement. Running "2008" as the calendar year and running it as September–November produce very different numbers. Both are honest; neither is complete.

In the data

A replay needs a price history for every holding across the episode, and today's portfolio often holds things that did not exist then. Ask for the minimum-volatility fund's prices from September 2008 and the first one you get is this:

The fund launched three years after the crash. A replay of autumn 2008 has no move to apply to it, and a spreadsheet that fills the gap with zero reports the holding as unharmed when the truth is unknown. Before reading a replay's total, check which holdings were actually old enough to be in it, and say what stood in for the rest.

Try it now

1. Both sleeves are below over their full histories, drawn on adjusted closes so the figures are total returns. Navigate both to 2008 and Measure the calendar year on each.

2. Do it twice more, over 2020 and over 2022. You now have three pairs of annual moves. Apply each pair to a 60/40 portfolio of $100,000, compute all three scenario losses, and rank them.
3. Note which year the bond sleeve helped and which year it hurt. Write one sentence describing that difference — as a fact about those years, not a claim about any future year.