WHAT THE NUMBERS COVER

Options trades from OPRA, the consolidated US options tape. Paid plans see it in real time. The free plan and the public pages show it on a delay, and every screen that shows delayed data says how long the delay is.

Every figure is labelled as what it is: observed off the tape, inferred from where a print landed against the quote, or modelled. Nothing here is investment advice, and nothing tells you what to trade.

Every total and every ranking on this product is computed over the prints we ingested, not over everything that traded. Prints under $10,000 of premium are filtered out when the tape is ingested and never reach the store, so they are not in any figure here.

That is most prints, and a smaller but still substantial share of premium.

The share varies by symbol and is not known for most of them. So a ranking here is partly a property of that filter rather than of the market: a symbol whose flow is mostly small prints is under-represented against one whose flow is mostly large, and we cannot tell you by how much.

Index options (SPX, SPXW, VIX, NDX, RUT and other index roots) are on the tape but excluded from totals, rankings and scores.

Cash-settled index options — SPX, SPXW, VIX, NDX, RUT and other index roots — are excluded from every total, ranking and score. Exchanges occasionally cancel ("bust") a trade after it prints. Our data feed doesn't yet tell us when that happens, and index options trade in such large notional sizes that a single cancelled print could outweigh a whole session of genuine flow. Leaving them out keeps one bad print from distorting everything else. You can still see every index print on the flow tape and each contract's page. We'll revisit this once we can detect and correct cancelled trades.

A lean is our reading of which side started a trade, taken from where the print landed against the quote at that moment. Exchanges do not publish who initiated a trade, so every lean is an inference, and we say how strong it is.

High confidence means the print traded at or through the offer (buy-side) or the bid (sell-side): somebody took the price on the other side. Medium confidence means it traded inside the spread, clearly nearer one side than the other but without crossing it — shown as "BUY-SIDE LEAN · MEDIUM CONFIDENCE" or "SELL-SIDE LEAN · MEDIUM CONFIDENCE". A print in the middle 30% of the spread, on a quote wider than its own midpoint, or forming one leg of a detected multi-leg trade gets no lean at all. Buy-side on a call, or sell-side on a put, reads bullish; the reverse reads bearish.

The two tiers are never added together. The Flow Bias line, every directional total and the direction filter count high confidence only; the medium figure is shown beside them, labelled as medium. Alerts fire on high confidence only, so an alert always matches the line you see on the chart, and only prints that actually crossed the quote can trigger one.

The lean is worked out from each print's recorded price, bid and ask every time you look. A file you have already downloaded, or an alert already sent, is never changed.

Prints before 2026-09-16 were additionally filtered at 10 contracts; prints after that date were not. A range spanning that date therefore mixes two rules, and its excluded share sits somewhere between the two.

Stored history is never recomputed, so this is permanent rather than something that will be cleaned up later. Where a page shows a window that spans the change, it names both rules rather than quoting one number that describes neither.

Intraday pages show one session at a time, and the session they are showing is stated on screen rather than left to be read off a date column.

If our feed missed part of a session, the status icon beside the page title says so and names the missing window. A clean icon means the tape ran without a detected gap; it does not mean the market was quiet.

A CSV or JSON export repeats the qualification inside the file, as a comment or an envelope, because a file outlives the page it came from and may be read by somebody who never saw this one.