Every listed option in the United States prints on one of eighteen exchanges, and the trades and quotes from all of them are consolidated into a single national tape under OPRA. That tape is the market. Anything less is a sample of it.
Data vendors face a real cost problem here. The consolidated feed is expensive; a single exchange group’s own data product is not, and Cboe’s is the best of those — well documented, deep history, priced so an analytics shop can actually build on it. Nothing is wrong with that feed. The question is the denominator underneath the numbers built from it.
The split, measured
Cboe publishes its own market-share statistics. As of late August 2026, its four options exchanges combined printed roughly 26% of SPY options volume, 24% of QQQ, and 19% of IWM. Those shares drift month to month; the shape does not. Meanwhile SPX and VIX options are Cboe-proprietary products — 100% of that volume prints on one exchange by construction.
On SPX, a single-exchange feed is the market. On SPY it is a quarter of it — and the label on the dashboard rarely changes when the denominator does.
Cboe’s figures are its own, per ticker. The clearing house sees all of it. Every listed option in the United States clears through the OCC, and its per-exchange split is the whole market rather than one venue’s self-report. Measured there on 31 August 2026, Cboe’s exchanges were 98.18% of index option volume and 11.87% of equity volume — and the OCC counts ETFs like SPY, QQQ and IWM as equity.
One feed is almost the entire index tape and about a ninth of the equity tape. Same feed, same label, two completely different denominators.
That is why we log it daily instead of quoting it once. The OCC keeps a rolling two-year window, so an unlogged day is gone rather than late — and a share only means something against its own history.
Where the gap hides
Open interest clears centrally at the OCC, so an OI-built ladder — the walls, the big strikes — looks identical from either source. The flow-derived reads are where the sample shows: put/call ratios, volume at a strike, buy-versus-sell inference from open/close data. Built on one venue’s prints, each of those is measured on about a quarter of the ETF tape. And it is not a random quarter — exchanges compete on fee structures and order types, so each venue’s mix of participants skews in its own direction.
The question to ask
Ask any vendor two things: is the source OPRA-consolidated or single-exchange, and for each number, is the denominator the whole market or one venue? A shop doing SPX-first work on Cboe data is being exact — there the single feed covers everything. The same method quietly extended to SPY, QQQ and IWM keeps the name and loses three-quarters of the denominator.
The chain NoVo reads — volume and open interest per contract — is the consolidated view, and the sweeps and blocks on the Analyst dashboard are computed in-house off the consolidated print tape. Where any number on the map is measured on less than the whole, the sample size is printed beside it. Same rule as everywhere else: the method is public, so the denominator is too.