Does every trade happen on an exchange?
Surprise: no. A large share of trading — in the US, routinely around a third or more of stock volume — happens off the public exchanges. Knowing where else trades live explains data quirks you'll meet and prices you'll occasionally question.
The off-exchange world
- OTC (over-the-counter) markets. A dealer network rather than a central venue. Two very different neighborhoods share the name: the professional OTC world of bonds and currencies (enormous, institutional), and OTC equity lists of small companies that don't meet exchange listing standards — thinner disclosure, wider spreads, and the natural habitat of the "penny stock" stories that end badly. For a beginner, the practical takeaway: exchange listing is a quality filter; its absence is information.
- Dark pools. Private, regulated venues where institutions match large orders without showing them publicly first. The name sounds sinister; the purpose is mundane — a pension fund selling a million shares doesn't want to announce it and move the price against its own savers mid-order. Completed trades still get reported publicly.
- Internalizers. Brokers routing retail orders to wholesale market makers who execute them directly — a normal fate for small orders in several markets, still bound by best-execution duties you met in Course 1.
Why fragmentation doesn't (usually) hurt you
Dozens of venues trade the same share simultaneously — yet regulation stitches them together: brokers must honor the best available prices across the public landscape, and consolidated reporting collects all trades into the tape your chart draws. Fragmented plumbing, unified price.
The data-literacy hook
When volume numbers differ between sources, or a "last price" briefly disagrees between apps, venue fragmentation is often the boring explanation. You'll formalize this instinct in the next course's data unit.
In the data
Fragmentation is visible trade by trade. Every single print on the consolidated tape carries a mark for where it happened, and trades done away from the public exchanges — in dark pools and at wholesalers — are reported under their own off-exchange mark. We added up one full regular session of Apple, 25 September 2026 from 09:30 to 16:00 New York time, print by print (measured 28 September 2026):
| shares | prints | |
|---|---|---|
| every print in the session | 22,608,804 | 589,441 |
| prints reported off-exchange | 10,919,480 | 56.9% of all prints |
| off-exchange plus three other public exchanges (NYSE Arca, Cboe EDGX, Cboe BZX) | 15,128,657 | |
| average size of an off-exchange print | 32.6 shares | |
| average size of any other print | 46.0 shares |
The daily volume you see on a chart is the first row: every venue added together. The rows under it are the split that one number hides.
Try it now
- Compare the latest session's volume for a company across two sources — small discrepancies are consolidation timing, not errors. Apple's from the end-of-day record is below; open the same session in the Terminal, or in any app you already use, and set the two numbers side by side.
Open AAPL.US in the EODHD Terminal
- Measure the claim in the first paragraph instead of taking it. In the table above, divide the off-exchange shares by the session's shares and set the answer against the "around a third or more" this lesson opened with.
- Resist the shortcut of counting everything that is not the listing venue. NYSE Arca and the two Cboe exchanges are public exchanges, lit and quoted, just not the one the company listed on. Divide the third row's shares by the session's and you get the share you would report if you counted them as off-exchange: about two thirds instead of about half, which is how a plausible measurement becomes a wrong one.
- Now read the print column. Off-exchange prints are a larger share of the count than of the shares, because a trade routed off-exchange is typically smaller, as the two averages show. Two honest numbers, two different questions, and only one of them answers "how much of the volume".
- Try to put an OTC-traded name's spread next to an exchange-listed one, and meet the filter in the data itself. We asked for a delayed two-sided quote on Nestlé's OTC receipt, NSRGY, and on Apple together on 28 September 2026: Apple came back with its bid and ask, and the OTC name with nothing at all. The quote service that carries both sides covers exchange-listed shares, so on this data an OTC spread is not wide — it is absent. Open the receipt in the Terminal and see what it does show.
Open NSRGY.US in the EODHD Terminal
- Say the quality-filter sentence once: "where something trades tells me something about it."