Scenario analysis as Apex practises it: a small set of named scenarios, each active or not, each with what would confirm and what would kill it — and worked examples from 2026.
A scenario is not a forecast. It is a named configuration of the world — "rates up", "geopolitical escalation", "flight to safety", "capital leaves both stocks and bonds" — with a status (active, not active, invalidated) and two lists: what would confirm it and what would kill it. Apex holds six at a time. On 19 September 2026 two were active (rates up, geopolitical escalation), three not active, and one — capital leaving both — invalidated.
Ask the channel first. If headline inflation rises but the real rate stays near its 99th percentile, the bond market is saying the rise will be met, not accommodated. History inside Apex's record: gold falls about 1 % for every 10 bp the real rate rises on jobs days; Bitcoin's correlation with the real rate is negative; the sector spread moves toward cash-generating cyclicals. The scenario therefore favours energy and cash flow over duration and gold — and it is invalidated the day the 10-year falls back through 5 %, because then the real rate is no longer the brake.
This was September 2026. Strategic pressure elevated (six of nine signals active), yet semiconductors led and the defensive-minus-cyclical spread was negative. Two readings compete: capital is ignoring the pressure, or capital is right that the pressure is contained. Apex's rule is to name the state without resolving it — "selective risk-taking under pressure" — and to list what would resolve it: a defensive turn in the spread, the crypto cycle flipping bear, the pulse collapsing. Until one of those prints, the honest scenario is unresolved, and the honest position is smaller.
A prediction is graded once, at the end, and usually by the person who made it. A scenario is graded continuously by its own kill conditions, and anyone can check them. Apex logs every scenario state every morning and records what followed; the daily pages at the Apex day are that log, and the research pages are what survived it.
Alongside the scenarios, Apex finds the closest historical month to the present configuration (in September 2026: September 2018, at 0.48 sigma, against 2.86 sigma for the dotcom peak) and reports how such months resolved — continuation, correction or reset — as frequencies, not as a call. History is a prior, not a prediction.
All numbers on this page were published in Apex’s own record at the time; they describe the past under stated conditions and promise nothing about the future. Research · Learn · The Apex day