Hypotheses & scenarios
Forecasts built from named assumptions you can see, challenge, and reuse.
In Aqmen, a forecast is never a typed-in number. It's the output of quantitative hypotheses (explicit, named assumptions about how drivers evolve) dialled per scenario. This is the mechanic that turns "what's driving 2030?" from a cell-reference hunt into a one-line answer.
Hypotheses
A hypothesis is a per-period growth impact on a driver: "pricing hardens 2.0pp per year from 2026", "enterprise penetration accelerates post-regulation". Hypotheses attach to driver cells and their impacts accumulate: several hypotheses can act on the same driver, and each one's contribution stays visible.
Because hypotheses are named objects rather than buried formulas:
- they can be challenged individually in review: accept the structure, debate the assumption;
- they can be reused across the scenarios of an analysis: one pricing thesis, dialled differently in the base and downside cases;
- they document themselves: the forecast's logic reads as a list of claims, not a wall of cell references.
Scenarios
Scenarios (base, upside, downside, or whatever the deal calls for) are dial settings over an analysis's hypotheses. A scenario doesn't fork the model; it turns each hypothesis up, down, or off. Switching scenarios recomputes the whole analysis instantly, so comparing outcomes is a toggle, not a rebuild.
This is also why scenario comparisons stay honest: within an analysis, every scenario shares one structure and one set of historicals. The only differences between base and downside are the assumption dials: exactly the things worth debating.
The agent way
Agents work with the same objects: ask for "a downside scenario where pricing stays flat and adoption slips two years" and the agent creates or adjusts the hypotheses, dials them into the scenario, and reports what changed. Reviewable, like everything else.