Derived data

How it works

Derived data is the third licensing axis, after who reads the number and where the number goes. Nasdaq's data policies define it as information generated in whole or in part from exchange information such that what is generated cannot be reverse engineered to recreate the exchange information, or used to create other data recognisable as a reasonable substitute for it. The named examples are instruments and aggregates rather than analytics: swaps, swaptions, spread bets, contracts for difference, volume-only data, and volume-weighted price.

Read the definition twice, because it is a fence rather than a category. The question is not "did I do arithmetic on it". It is whether the output either reconstructs the input or stands in for it. A last price rounded, delayed by a second, republished as a "fair value" or wrapped in a single-symbol index is a substitute; a firm-wide risk number computed from ten thousand quotes is not. The calculation's sophistication is irrelevant, which is why a derived-data question cannot be answered by describing your formula.

Where the line actually falls

Nasdaq's policy puts three notes under the definition, and between them they draw the line more usefully than the definition does.

• Price in, underlying rate out. Derived data that contains price data is generally fee liable at the underlying product's rates. So the derived category is not a discount on a number that still carries a price in it.
• One-for-one trackers are the underlying. Single-security instruments the exchange supports, or instruments that track a financial instrument one for one, are fee liable at the underlying rates. A wrapper is not a transformation.
• Many securities in is different. For equities, creating multi-security information such as an index is exempt from subscriber fees — but the distributor must report the non-display devices involved. The fee does not vanish; it changes shape, from per-subscriber to non-display use.

The scaling is by audience, and this is the part that surprises people building a product. Nasdaq's US Basic derived data created and distributed externally only to non-professional subscribers is fee liable as part of a specific derived data fee; derived data distributed to professional subscribers is fee liable at the regular per-subscriber or per-user rate. In other words the same calculation is priced by how many end customers see it and which category they fall in — the professional subscriber question again, one layer up. And a firm receiving a real-time feed still owes at least the internal distributor fee for receiving it with derived information, so nothing here is a route to taking a feed for nothing.

Index values are the sharpest case, and the tape's own delayed data policy makes it explicit from the other direction: the plan participants extend no preferential treatment to index information, the same contract requirements apply to delayed index values as to real-time ones, and index information is excluded from the data elements that get the cheap delayed treatment. An index built on a venue's prices can carry a licence of its own from the index provider on top of that — the assembled membership history is a licensed product in its own right, which historical index constituents covers.

Why it matters here

Three practical readings, for anybody deciding what a screener or a chart may output.

A screen output is not automatically yours. The ranking, the score and the indicator column are derived from somebody's prices, and if that output leaves your organisation it is both a derived data question and a redistribution question. The second one is usually the expensive answer, and the first does not neutralise it.

"Computed locally" is a technical fact, not a licensing one. Running the calculation in your own process changes nothing about the prices that went in. This is why vendors that ship indicator endpoints — Twelve Data and Alpha Vantage among the cards here — have terms about what you may publish, not just rate limits.

Ask the vendor which axis they have already paid for. A vendor holding a distributor licence may or may not hold the derived and index rights beside it, and it is not a question the pricing page answers. A vendor that itemises exchange fees, such as Databento, at least lets you see which products are on the account.

What real-time market data costs has the fee stack these categories sit in, and entitlement explains the per-subscriber counting the derived rules keep referring back to.