Stress testing asks a deliberately uncomfortable question: what would this portfolio do if something bad happened?
It is not forecasting. A scenario does not need to be likely to be worth examining — it needs to be specified well enough that its consequences can be traced through the holdings.
Two kinds of exercise
Historical stress tests apply a past episode to the current portfolio: what would happen if conditions resembling a named crisis recurred. Their strength is that the episode actually happened, so the relationships between assets during it are known rather than assumed. Their weakness is that the next crisis will differ.
Hypothetical scenarios construct a future that has not occurred: a specified move in rates, a specified change in credit spreads, a sector-specific shock. Their strength is that they can be tailored to the portfolio's actual exposures. Their weakness is that the relationships between assets must be assumed, and assumptions about correlation are exactly what tends to fail under stress.
Both are useful, for different reasons. A serious process uses both.
Building a scenario that tests anything
A scenario is only useful if it is specific about transmission. "A recession" is not a scenario. A scenario says:
- What moves, and by how much. Interest rates, credit spreads, equity indices, commodity prices, currencies, liquidity.
- In what order. A shock that begins in credit and spreads to equities has different implications from the reverse.
- Over what period. A fast shock and a slow grind through the same total move affect different portfolios differently — the first may force sales, the second may not.
- How the holdings respond. Not through a single beta, but position by position: whose revenues fall, whose spreads widen, whose funding is withdrawn, whose collateral is revalued.
The last point is where most exercises stop short. Applying an index move to the portfolio total assumes the portfolio behaves like the index, which is exactly what the exercise is meant to check.
What to look for in the output
The point is not the headline loss figure. It is the mechanism:
- Which positions drive the loss, and whether they were expected to.
- Whether the losses are correlated in the scenario, and whether that correlation differs from the recent past.
- What would have to be sold, and whether it could be sold at the assumed price.
- What the forced actions would be. A margin call, a redemption, or a funding withdrawal converts a drawdown into a permanent loss.
- What the second-order effects are. Collateral, covenants, counterparty exposure.
The uncomfortable scenario
The scenario worth running most is the one that is least likely and most damaging. It is also the one easiest to leave out, because it feels unrealistic and because the preparation, if it were taken seriously, would require unwinding attractive positions.
Eonvera's Scenario & Causal Intelligence module applies probabilistic research, Monte Carlo simulation, stress testing and multi-scenario analysis to examine portfolio outcomes and capital allocation choices under alternative market paths. The purpose is not to predict the path but to ensure the portfolio can survive more than one of them.
Scenario work is a process, not a report
A scenario analysis that is produced once and filed is a document. The value comes from the cycle:
- Define the scenario.
- Trace it through the actual holdings.
- Identify the vulnerabilities it reveals.
- Change something — reduce the position, add a hedge, arrange liquidity, or consciously accept the risk.
- Repeat as the portfolio changes.
Step four is what distinguishes the exercise from an academic one.