How we count a power trading market

Every number on this site comes from a register we build and maintain ourselves. This is the method, including where it is imprecise and why we publish that alongside the figures.

The unit is the person in the seat. We count the people who trade, price, forecast, optimise, build and lead in European wholesale power, held as five role families, each within its own country. A market is counted role by role: not how many people work at trading firms, which counts receptionists, but how many people do this specific job in this specific country. That definition is the whole method. Every downstream number inherits it.

Inclusion is decided on evidence, not keywords. Titles in this market are unreliable, the same job carries a dozen names across firms and languages, so a person enters the register on what their work demonstrably is, and ambiguous cases are held out until the evidence settles them. This is slow. It is also the difference between a census and a keyword search, and the errors it prevents are exactly the errors that would make the numbers useless: the "energy trader" who trades oil, the "analyst" who runs a book, the "developer" who has never touched a trading system.

Every figure carries the date it was counted. Markets move, people move, and a number without a date decays silently. Dating the figures is how we keep ourselves honest about staleness, and it is why numbers on this site change rather than accumulate.

Where it is imprecise, we say so. A small share of people cannot yet be placed in a single country with confidence, and they are held out of regional figures rather than guessed at, which is why shares can sum to slightly less than one hundred. Some pools are too small to say anything useful about, and the threshold is fixed rather than judged case by case: below [n] people, we publish the fact of the small pool instead of a number, because a figure built on a dozen people is a coin toss wearing a percentage sign.

Why publish the limitations at all? Because the register's value depends entirely on being believed, and the fastest way to lose that is to publish numbers that cannot survive their own definition being probed. Every figure here can be asked "counted how?" and answered in one sentence. That property is the product. The counting is just how we get it.