OPRA

How it works

OPRA describes itself as a securities information processor registered under Section 11A(b) of the Securities Exchange Act of 1934, disseminating consolidated last sale and quotation information originating from the national securities exchanges approved by the SEC to provide markets in exchange-traded options. Its plan is a limited liability company agreement, and the participants are the exchange families that list options: BOX, Cboe, MEMX, MIAX, Nasdaq and NYSE, which between them operate most of a dozen and a half separate options exchanges.

The important word is one. In US equities there is a genuine choice between the plan-run consolidated tape and an exchange's own proprietary feed, two orders of magnitude apart in price, and a card can be cheap because it chose the first. In listed options the quote and the last sale come from one processor under one plan, so there is no cheaper consolidated alternative to fall back on and no per-venue shopping to do. Options data prices as an asset class because it is administered as one.

What the schedule actually says

Its shape is worth knowing, because every options card's pricing page is downstream of it. On the schedule dated February 2026:

• a professional subscriber with an agreement directly with OPRA pays a device-based fee of 31.50 dollars per display device per month;
• non-professional subscriber fees are paid by the vendor and tier down with volume, from 1.25 dollars per non-professional subscriber to 60 cents above 500,000 of them;
• a vendor may instead elect a usage-based fee, charged at three quarters of a cent per "quote packet" or 3 cents per "options chain", subject to a stated monthly maximum;
• redistribution is 1,500 dollars a month, or 650 for a query service only, and it does not apply to a vendor whose redistribution is limited solely to historical data;
• a professional receiving OPRA data indirectly, as a data feed transmission from a vendor, pays 600 dollars a month, and that fee applies where the feed is taken solely for non-display use.

Two things in that list explain most of what readers meet in the wild. The unit of the usage-based fee is the options chain — the thing an application actually requests, one underlier's strikes and expiries at once — which is why chain endpoints are rationed and metered in ways an equity quote endpoint is not. And the historical carve-out from the redistribution fee is why a vendor will happily sell you a year of prints as files while quoting a live feed at a different order of magnitude.

Why it is the extreme case

OPRA is the largest message source in this market, by a distance nobody disputes. The plan's revised capacity projections of 15 September 2025 project 13.575 billion messages a day effective July 2026, rising to 14.964 billion by July 2027 — and that is for one of two redundant streams. What that does to storage and query design is the subject of tick data and storing tick data; what it does to products is simpler. Nobody sells a hobbyist a full historical options feed. They sell files, by underlier and by date, with the NBBO attached at trade time.

Why it matters here

For the options cards in this catalogue, OPRA is the reason they all look alike in the ways that matter. Cboe DataShop, AlgoSeek and Tick Data sell history as files because that is the cheap side of the schedule. ORATS and Market Chameleon sell derived values — implied volatility surfaces, skew, earnings-move statistics — partly because computation is theirs to keep while the underlying quotes are not.

Three questions follow for any options product here. Is the live quote OPRA data, and therefore carrying the plan's per-user or per-chain fee? Is what you are given historical, and therefore outside the redistribution line? And if you intend to show it to anyone else, or feed it to a program, which of the two expensive words applies? The options data collection is the listing; what real-time market data costs is the arithmetic; and how to get an options chain is where the metering above turns into a rate limit somebody actually hits.