Explainers
What a crisis parallel is, and is not
A parallel says which past episodes looked like today at the same depth and speed. It describes the past. It does not say what happens next, and the hub never presents it that way.
When a market has fallen, say, 18% in two months, it is natural to ask which past falls looked like this one. The crisis library answers that question with a ranked list and a reason for each entry.
How the ranking works. Each past crisis in the library has a few plain facts: how deep the index fell, how many months the fall took, which markets were at the centre, and what kind of episode it was. The ranking compares those facts with the situation you describe and lists the closest episodes first, with the distance shown. No hidden model, no prediction.
What you can take from it. The range of outcomes after similar starts. For example, falls of about a third have been followed by a new high after five months (2020) and after twenty-five years (1929). That range is the honest answer.
What you cannot take from it. A forecast. The library has a few dozen episodes, each with its own cause; two episodes that look the same at month two have ended very differently. A parallel is context for a decision you make with your own limits in place.
Where it comes from. The case files in the research section, one per crisis, with sources and a confidence note. Figures from memory are marked as approximate there.