Free editorial articles · Inquiries only
A persistent myth treats AI in investment as an oracle that forecasts markets. In practice, the useful work is narrower and more honest: building scenario sets and running portfolio stress tests that show how a collection of holdings responds to defined shocks. We publish free articles explaining how that machinery actually works.
Myth-busting
Each myth below is something we hear often. The correction is what our articles actually walk through.
Myth 01Scenario analysis does not forecast the next crisis. It takes a named shock — a 200-basis-point rate rise, a liquidity freeze, a supply-chain break — and asks how the portfolio moves under it. The output is a conditional loss figure, not a prophecy.
See how scenarios are framed →
Myth 02Any multi-asset portfolio carries factor exposures — equity beta, credit spread, duration, currency. A stress test propagates a shock through those factors. We explain the same propagation in plain language for holdings readers actually own.
Read the propagation notes →
Myth 03Without a scenario hypothesis, more data mostly adds noise. Our articles describe how a small, well-chosen scenario set keeps the analysis interpretable — and why a hundred extra variables rarely change the loss ranking.
Compare scenario sets →
A scenario is a disciplined question asked of a portfolio, not a forecast of the market.
Editorial position — Horizon Module ResearchProcess
The same five moves appear across most scenario and stress-testing work. Our articles follow them in order.
List the holdings, then translate them into factor exposures — equity beta, credit spread, duration, currency and liquidity. The exposure map is what the shocks act on.
Combine a historical replay (a past crisis applied to today’s book), a hypothetical shock (a defined rate or spread move) and a reverse stress test (what shock would have to happen to breach a loss limit).
Apply each scenario’s risk-driver moves through the exposure map and revalue the positions. This is where AI tooling helps — running many revaluations quickly, not predicting the drivers themselves.
Collect the revalued results into a loss distribution per scenario. The useful output is the ranking and the concentration of losses — which holdings cluster in the worst outcomes.
Record what each scenario assumes, what it ignores and where the model breaks. The article set always states this plainly: no forecast, no personalised advice, no execution.
Honest scope
We would rather state the limits plainly than oversell the role of AI in investment.
If you want to understand how a multi-asset book responds to named scenarios — rate shocks, spread widening, currency breaks — the articles walk through the method without selling you anything.
The scenario-construction steps, exposure mapping and loss-distribution reading are explained end to end, with the assumptions stated rather than hidden behind a black box.
We do not offer investment advice, portfolio management, trade execution, deposits or subscriptions. If that is what you need, an inquiry will be answered with that limit stated plainly.
Make an inquiry →