Where Market Data Comes From: Exchanges, Consolidated Feeds, and Aggregators
5 steps · one page
In short
Every price on every screen began life in the same place: a trade or a quote inside a trading venue's matching engine.
Everything after that moment is distribution — a supply chain of consolidation, licensing, redistribution, and display that ends at your quote panel, and whose structure explains most of the everyday puzzles this pillar keeps meeting: why two screens disagree, why free data is delayed, why reported volume differs between sources, and why "the price" is a simpler idea than the pipeline that delivers it. This article maps the chain in three stages — the origin (venues), the consolidation layer (public tapes and proprietary feeds, which differ sharply between the US and Europe), and the vendor layer (the aggregators and platforms that actually reach retail screens) — reported as industry structure, with no vendor recommended and the live policy arguments reported as arguments.
The origin: matching engines, trades, and quotes
A modern exchange is, at its core, a matching engine: it maintains an order book per instrument and, when a buy meets a sell, produces a trade print — price, size, timestamp, venue. Alongside executed trades, the engine continuously emits quote updates — the best bid and ask (and, deeper, the whole book, which is the order-book data article's subject). These two streams — trades and quotes — are the raw material of all market data, and two structural facts about the origin shape everything downstream. First, trading is fragmented: the same security trades simultaneously on multiple exchanges and off-exchange venues (alternative trading systems, and trades executed by wholesalers away from exchanges, which in the US must be reported promptly to a trade-reporting facility so they appear in the public record). No single venue sees the whole market — consolidation has to be manufactured. Second, the data is the venue's product: as the latency article covered, exchanges monetise these streams, which is why every later stage of the chain is shaped by licensing — who may see what, how fresh, at what price, under which professional/non-professional and display/non-display terms. A practical corollary lands immediately: reported volume depends on which of these origins a source counts. A number that includes off-exchange prints differs from one that counts a single exchange — which resolves, mechanically, the volume-convention caveat the quote-anatomy article flagged: neither number is wrong; they draw the boundary of "the market" differently.
The consolidation layer: public tapes, proprietary feeds — and two regions' answers
Because trading is fragmented, someone must stitch the venues together. The US answer is institutionalised: regulated consolidated feeds — securities information processors (SIPs) operating under joint industry plans — collect every venue's trades and best quotes and publish a single public stream: the consolidated tape, from which the NBBO (national best bid and offer) is computed, the reference point around which US retail order handling and execution-quality rules are built. In parallel, exchanges sell proprietary feeds — faster and deeper than the SIP (full order books, auction imbalances) — creating the two-speed structure that professional participants pay for and public data trails: a long-running policy controversy (is the public tape fast and rich enough? are proprietary-feed prices fair?) that produced years of rulemaking, litigation, and an infrastructure-modernisation program, reported here as the live argument it remains. The European answer has historically been: no single tape. Post-MiFID fragmentation spread trading across national exchanges, pan-European venues, and off-exchange reporting channels, with publication through approved publication arrangements — but no consolidated public stream; anyone wanting a whole-market view has bought and merged feeds vendor by vendor, which is why pan-European "consolidated" views have effectively been a commercial product rather than a public utility. That is now changing by regulation: under the revised MiFIR framework, the EU is standing up consolidated tapes per asset class through ESMA-selected tape providers — and as of August 2026 the build-out stands at providers selected for all three asset classes (bonds in July 2025, shares and ETFs in December 2025, OTC derivatives in July 2026), the equities provider authorised by ESMA, and no tape stream live yet. The regional contrast is itself the literacy: "the price" a US screen shows is usually anchored to a regulated public consolidation; a European screen's price is anchored to whichever venue or vendor-merge sits behind it — one more honest reason two screens disagree.
The vendor layer: aggregators, platforms, and the last mile
Between the tapes and your eyes sits the aggregation industry. Market-data vendors and aggregators license feeds from many venues and consolidators, normalise them (symbols mapped across identifier systems, timestamps aligned, corporate actions applied per adjustment conventions), and redistribute them onward — to professional terminals, to brokers and retail platforms, to websites and apps, and increasingly through APIs that the platforms-and-APIs article surveys. The categories range from the full-service professional terminal vendors through specialised feed consolidators to retail-oriented data services — described here as categories, with none recommended. Three structural facts about this last mile complete the map. Licensing follows the data downstream: a platform showing you real-time prices holds redistribution licences and reports entitled users to the venues — the administrative machinery that makes "live data" a per-user product and delayed data the free default. Normalisation is editorial: vendors make choices — which venues' prints to include, how to treat off-exchange volume, which adjustment conventions to apply, how to map identifiers — so two honest vendors can publish different numbers for "the same" field; sophisticated consumers read vendor methodology notes the way this portal keeps recommending readers check chart conventions. The chain adds latency at every hop: origin → consolidator → vendor → platform → screen, each stage adding processing and transport time — which is why even "real-time" is a spectrum, and why the professional end of the market spends heavily to shorten it, per the HFT article's documented arms race. End to end: a trade prints in a matching engine; consolidators merge it into a tape; vendors normalise and redistribute it; platforms display it under licence, at the freshness tier the viewer's entitlement allows. That is the whole industry in one sentence — and every screen you will ever read sits somewhere specific on that map.
Worked example
Worked example (fictional). At 15:02:10.412, fictional stock ARVEN prints a trade at $52.30 on one exchange. By 15:02:10.6, a professional desk watching the venue's proprietary feed sees it. At ~15:02:11, the consolidated stream carries it and platforms with real-time entitlements — including a broker app and a Pro-tier analytics screen — display $52.30. A free website shows the print at 15:17, when the delay embargo lapses. Meanwhile a second site reports the day's volume 8% higher than the first — it counts off-exchange prints the first excludes. Every screen was honest; they sat at different points of the same supply chain. (All names, tickers, and figures fictional.)
Frequently asked
5 questions
Who actually produces stock market data?
Trading venues — their matching engines emit trade prints and quote updates as trading happens, and off-exchange trades enter the public record through reporting facilities. Everything else in the industry — tapes, vendors, platforms — is consolidation and distribution of those origin streams.
What is the consolidated tape?
A single public stream merging all venues' trades and best quotes for a market. The US operates regulated consolidated feeds (SIPs) from which the national best bid and offer is computed; Europe historically had none — whole-market views were assembled commercially — and is now standing up consolidated tapes per asset class under MiFIR reforms: as of August 2026, providers are selected for all three asset classes, the equities provider is authorised, and no stream is live yet.
Why do different sites report different volume for the same stock?
They draw the market's boundary differently: some count one exchange, some all exchanges, some include off-exchange prints reported through trade-reporting channels. The numbers differ because the definitions differ — a supply-chain fact, not an error.
What do market-data vendors actually add?
Consolidation and normalisation: licensing feeds from many venues, mapping identifiers, aligning timestamps, applying adjustment conventions, and redistributing the result to terminals, platforms, and APIs. Their editorial choices — venues included, conventions applied — are why two honest vendors can publish slightly different numbers for the same field.
Is the public data worse than what professionals see?
It is slower and shallower, by structure: proprietary feeds carry full order books and arrive faster than public consolidated streams, and the fairness of that two-speed arrangement is a genuine, long-running policy argument — with rulemaking and litigation on the record — that this portal reports as unresolved rather than referees.
References
Educational and informational only — not investment advice, a recommendation, or an offer to buy or sell any security. Investing involves risk, including the possible loss of principal. Worked examples use fictional companies and figures.