Wave Foundation — The Variables as Time Series

An ROE of twenty percent is just a number. It can be the top of a falling arc, the bottom of a rising one, or the midpoint of a decade of stagnation — the same figure, three completely different companies. But the path there — from sixteen percent to twenty, with the gross margin widening and the debt-to-equity ratio narrowing — is a story.

The story tells what the number is about to become; the number only tells where it is. The whole dividing line between ordinary fundamental analysis and the Wave Foundation lies in that sentence: we stop asking how high and start asking since when and in which direction.

Think of two companies with identical AKM1 profiles: the same score on growth, profitability, stability. The still image says they are interchangeable. But in the first the margins widen with every report while the debt narrows; in the second the points have been reached on a downward staircase — growth bought with dilution, the margin saved by one-off items.

After ten reports the two companies have swapped places, and after fifty they have swapped category. No single scoring can see that. The time series sees it immediately. It is not a refinement of the model — it is the very claim: the movement of the variables carries information that the levels do not carry.

An important mark before we go on, because it shapes the whole course. The Wave Foundation engine classifies every variable on every horizon — but the engine's exact classification rules are the lab's internal know-how. What we teach in this course is something else and, we claim, more valuable: how you read and interpret the wave classes, how you build the hierarchy from cell to whole, and how you keep the honesty alive when the material is thin.

So you will not find thresholds to memorize in these chapters. You will find a language for movement — and the language is what lets you use the tool, whatever the engine is called tomorrow.

Revenue growth is driven by product cycles and the order situation — it answers the world quickly and is noisy by nature. The gross margin is driven by pricing, the cost base and currencies — it moves more slowly but with weight. The debt-to-equity ratio is not driven by the market at all but by management's capital decisions: a share issue, a sold plant, a new loan — and suddenly the series has jumped.

Patents and the brand are driven by strategy measured in years, not quarters. The valuation variables, P/S and P/B and EV/EBITDA, are really two series in one — the company's figures and the market's price — and therefore breathe with the market's mood. Twenty variables, twenty engines.

The consequence is that a company's fundamentals never have a single state. When you say the company is doing well you are making a kind of average of twenty stories — some improving, some deteriorating, some standing still — and the average hides the spread. The spread is not noise; it is the content.

A company whose profitability widens while growth corrects is on its way to becoming a different company than it was. The matrix's twenty rows make that transformation visible row by row, instead of blending it into a single feeling of health.

Where do the series come from? From the report stream. The quarterly reports are the pulse beats of the series — each beat leaves one value per variable, and the values in sequence form the history the engine reads. The material is fetched from a public data source, and where an AKM1 value can be computed from the material the level is shown as a score 0–5 beside the row; where it cannot be computed deterministically a dash is shown.

It is the same honesty on the level side as on the wave side: compute where possible, stay silent where not. The length of the history also varies between companies — a company with a short report history has a short series, however interesting it is — and that leads us straight into the horizons and the world of the unset class, which we will come to.