For writers & researchers

The unusual options activity board: the highest-volume US option contracts of a session among the 40 underlyings the board scans — at most 30 rows, ranked by single-contract volume, each carrying ticker, contract type, strike, expiry, volume, open interest, the volume/open-interest ratio and implied volatility. Only contracts that actually traded are eligible — a zero-volume strike is never listed.

This is a fixed scan list, not the whole US options market. Every 30 minutes the full option chain of each of these 40 names — liquid mega-caps, heavily-traded retail names and major index and sector ETFs — is pulled, and the highest-volume contracts across all of them become the board, capped at 30 rows. A contract on any other underlying will not appear, no matter how much volume it trades — not on the board, not in the day JSON, not in the CSV. If you need a superlative, the defensible one is "the most-traded contract among these 40 underlyings", never "the most-traded US option contract". Names outside the list, and quiet names inside it, are simply out of frame: an earnings-day surge in an unscanned ticker leaves no trace here.

Each archived ET trading day has its own permalink at https://www.tickwind.com/en/unusual/{YYYY-MM-DD} — for example today's board links its recent days. Those pages are server-rendered, so the rows are in the HTML a crawler or an answer engine sees, and each one carries a "Cite this data" block with the permalink, a paste-ready citation, CSV and JSON links and the licence line.

The archive is written once per ET weekday, at 16:38 ET, and stores whatever the newest completed board scan published at that moment. Scans run every 30 minutes, so that is normally a post-close scan carrying the session's full volumes; the feed's own timestamp travels with the day as data_at, so you can check which session the rows describe. A day is immutable once written: the first complete write wins and later scans never overwrite it, so a cited link keeps showing what it showed when you cited it. A restart that spans 16:38 runs one catch-up write on start-up; nothing else ever writes a day.

Days that have no snapshot stay empty rather than being filled in, and a missing day is not evidence the market was closed. Weekends never fire, and neither does any day before the archive began. On an ordinary trading weekday the write is also skipped — deliberately, rather than storing something stale or partial — when the board is empty at 16:38, when the newest scan is not from that ET day (a stalled scanner), or when the feed's own data day differs from that ET day. That last gate is how a weekday market holiday is caught: the scanner keeps running and re-serves the previous session's volumes under a fresh scan time, and the archive refuses it. A skipped day stays missing permanently — nothing is backfilled, estimated or reconstructed after the fact. Treat a 404 as "no snapshot for this day", never as "no trading on this day".

The endpoints behind the board are public — no key, no sign-up. A shared per-IP rate limit applies, so batch politely. Base URL https://api.tickwind.com.

• A malformed date is a 400; a well-formed date that was never archived is a 404 — the archive never invents a session.
• An archived day is served with a long immutable cache header; the live CSV (no ?date=) is no-store, because it changes with every board scan. Its filename carries the feed's own data day, falling back to the scan's ET day when the feed supplied no timestamp.
• The CSV carries the same values as the JSON, at full precision — no extra rounding, no re-derivation. (Formatting can differ: the CSV never uses exponent notation, JSON may.)
• data_at is the Cboe feed's own chain timestamp — the newest one among the chains that scan pulled — and is omitted, never zero-filled, when the feed supplied none. It is the feed's wall clock, so read it as a session stamp rather than an exact UTC instant.
• The live board is also available as JSON at GET /v1/options/unusual ({count, updated_at, contracts[]}) — the same rows the dated archive freezes each afternoon.

The reuse terms, exactly as they appear on every cite block:

Free to reuse with a link to tickwind.com.

The suggested citation each dated page offers, with that page's own ET day filled in:

Unusual options activity for {YYYY-MM-DD} — Tickwind (tickwind.com), Cboe 15-min delayed data.

And on the live board, which is not tied to one session:

Unusual options activity — Tickwind (tickwind.com), Cboe 15-min delayed data.

The citation sentence is the same English string in every locale — it is the citable artefact, not UI chrome.

Options figures come from Cboe and are delayed by roughly 15 minutes. State the source and the delay when you quote a number.

volume, oi and iv are the feed's own numbers, carried through unchanged. type, strike and expiry are decoded from the feed's own OCC contract symbol, and ticker is the underlying whose chain was requested. None of those are re-estimated — the site formats them for display and nothing more.

Exactly one row field is derived here: vol_oi, computed as volume ÷ oi. When open interest is 0 the ratio is undefined, and the field is written as exactly 0— a sentinel for "no prior open interest", not a small ratio.

That convention matters if you filter. Every row on the board traded at least one contract, so a positive open interest always yields a positive ratio: on this data vol_oi == 0 means precisely oi == 0. A brand-new strike trading 12,000 contracts against no prior open interest — the most extreme fresh-positioning case there is — therefore exports vol_oi 0, and a script that discards low ratios as noise discards exactly those rows. Read oi first, then the ratio.

For corrections, questions about methodology, or anything else, email [email protected]. The same address is on the contact page.