Gathering the data

Last time we counted the cost of complexity — the data every instrument demands, and the computation to reprice it — and ended on a promise: that keeping those inputs sound is a discipline of its own. This piece takes up the first half of it. Before anything can be priced, cleaned or checked, it has to be gathered, and the question that decides everything downstream is a plain one — who goes and gets the data.

Every risk number rests on market data and the question is who sources it, who keeps it current and who pays for it — the firm using the numbers, or someone who does it on their behalf.

Take one structured product — one line on a statement. To reprice it tonight, we need the price of every underlying it references; the yield curves to discount its future cash flows; the issuer's credit curve to weigh the chance it cannot pay; the exchange rates to carry all of it into the client's currency; and any corporate action — a dividend, a split, a merger — that affected an underlying since yesterday. These facts do not live in one place. They come from different houses, arrive in different formats, and update on different clocks. Several of them changed overnight.

That is one instrument. A real book holds thousands, of every kind and flavour. The task was never to look up a price. It is to gather a vast, shifting body of facts from many specialist sources and make them agree — every single day.

So we made the same decision here that we have made at every other turn in this series: when the work is laborious and the complexity real, we take it on ourselves. We source the data. We carry its cost. What reaches you is the finished number.

Where the market data comes from

We source the raw market data ourselves, from multiple top-tier providers. We do not pick one and make it do everything. A single source stretched across every asset class is excellent at some things and merely adequate at others, and the merely-adequate corners are precisely where risk hides. So we go to the specialist for each kind of fact — for example, one house for listed equities, another for fixed income timeseries, another for fixed income reference data, another for structured products reference data — and we hold the relationships, and the licences, with all of them. Between them they cover the full universe a wealth management book can contain.

Every day, thousands of timeseries are updated

Gathering is not a one-time event. Every day, more than 250,000 data series — credit ratings, rates, curves, holdings — are refreshed, so the picture you open in the morning reflects yesterday's markets.

The fast-moving facts update several times a day, following the markets around the world: Asia's close, then Europe. Knowing when not to fetch is part of fetching well.

Absorbing the mess

Here is the part that is invisible from outside, and is most of the actual work.

Every provider speaks its own dialect: its own formats, its own identifiers, its own conventions for the very same fact. Left as they arrive, the feeds do not fit together — you could not compute one coherent risk number across a portfolio assembled from them. So for each provider we run a dedicated service whose only job is to fetch that source and translate it into a single common form the whole engine can read. Many incompatible inputs in; one clean, unified dataset out.

Why we go through all this: so you never have to

Every part of this exists to take the whole problem off your desk. You never deal with a data provider — no feeds to choose, no formats to reconcile, no missing price to chase across vendors. You ask your question, and the data is already there. That ease of use is the point. It shows up in three concrete ways.

1. You onboard in a fraction of the time. Because the sourcing, the connections and the standardisation already exist, you connect and begin in a fraction of the time it would take to build any of this yourself. A firm doing it alone would negotiate with each provider, learn each format, build and maintain a separate pipeline for every one, and reconcile the lot into something a risk model can actually use — months of plumbing before a single number appears. You give us an identifier — the same code your custodian already uses — and we do the rest. The slow groundwork was finished before you arrived.
2. You pay no additional data license. Market data is sold under licence, and those licences are genuinely expensive. We hold them; you hold none. You are not buying and maintaining raw feeds — you are reading finished answers.
3. You pay a shared cost, not a solo one. Because we license each source once and serve the standardised result to everyone on the platform, that expense is mutualised across the whole client base rather than paid in full by each firm alone. A heavy fixed cost for one becomes a light shared one for all — a large part of why moving to Edgelab lowers the total cost of ownership instead of adding to it.

Behind a single number on your screen sits a daily, worldwide effort to find, fetch and reconcile the facts it rests on — providers, formats, schedules, failures and all. None of it reaches you, and that is the design. The complexity of gathering the data is real. We just keep it on our side of the line.