Docs and this page

Docs / DATASET

DATASET guide

Every finished SPX session as files on your own machine, with every field's book, unit and start date named.

The DATASET licence, monthly or yearly. Reading single sessions on the site stays free. The DATASET product page · Markdown copy

Why it matters

The archive is free to read one session at a time, on the site and in the terminal. Research over the whole of it needs the sessions as files you can compute on, and a clear rule for which book every number comes from.

DATASET answers one question: how do I run my own study over every session FirmTape has measured? It delivers the same day files the terminal reads, one a day, with the per-visitor limit on distinct days removed for your account.

Every session carries three books. The volume book is the conventional public object, built here from the tape; the open-interest book uses the morning's positions; the measured book is signed print by print from the tape and its quotes. Their magnitudes are never comparable, so every field names its book.

Where to find it

Start the licence on pricing. Once it is active, Account, Plans and billing shows how to download; every field is on the data dictionary.

You want	Go to
The licence	Pricing, DATASET
What a field means	The data dictionary
Which days exist	The day list the account page names
To show the numbers to other people	Licensing

DATASET: The whole SPX dealer-gamma archive, as files on your own machine.

Read it in 30 seconds

A field's suffix names its book: _volcp or _vol the volume book, _oi the open-interest book; a neutral name is the measured book.
The dictionary's since column says from which session a field exists; earlier sessions do not carry it.
Levels are SPX index points and times are New York time.
A day file covers one finished session; a session is never rewritten after its evening.
The calendar file carries each scheduled release and what usually followed its kind.

How to use it

The files

The day files and the dictionary#

Why: the same session files the terminal reads, as data for your own code. Where: with the licence active, Account, Plans and billing, How to download; every field on the data dictionary. Plan: The DATASET licence.

Fetch the day list to see which finished sessions exist.
Fetch each day's file with your account's token as a bearer header, from a script.
Look up each field on the data dictionary: its meaning, its unit, its book and the session it starts from.

The data dictionary: every field of the day files and every LAB column, grouped by book, with meaning, unit and start date

What it measures: A day file is one finished SPX session as the terminal reads it: minute series, per-strike ladders and the session's levels. The dictionary lists every field of the files and every LAB column, generated from the same source the LAB compiler reads, so the two never disagree.

How to read it:

A field's since column is the first session that carries it; a study over all sessions should start each field there.
Times are New York time; levels are SPX index points.

Limits: Finished sessions only. Your own research only: the files are not for showing to other people.

Developer reference: /docs/developers/dataset

Three books in every file#

Why: three readings of the same dealer positioning, side by side, and a rule that keeps them from being mixed. Where: the book tabs on the data dictionary. Plan: The DATASET licence; the dictionary is free to read.

Open the dictionary and pick VOLUME, MEASURED, OPEN INTEREST or NO BOOK.
Read the line under the book's name: how to spot its fields by suffix.
Compare magnitudes only inside one book.

The data dictionary's book tabs with the volume book open: cp times cumulative session volume, spotted by the suffix _volcp or _vol

What it measures: The volume book signs cumulative session volume by call or put (the conventional public object, built here from the tape); the open-interest book uses the morning's open interest; the measured book is signed print by print from the tape and its quotes. A field with no book (price, time, implied vol) is marked so.

How to read it:

_volcp and _vol are the volume book, _oi the open-interest book; a neutral payload name is the measured book.
The volume book is the default: it is what the chart draws when no book is named.
Two books disagreeing on the regime is expected; they measure different things.

Limits: Magnitudes are never comparable across books.

A new session every evening#

Why: the archive grows by one session a day, and your copy can grow with it. Where: the day list, the evening after each session. Plan: The DATASET licence.

After the close, fetch the day list again.
Fetch the new day's file; earlier days never change.
Fetch the calendar file too for the scheduled releases and what usually followed each kind.

What it measures: A finished session is written the evening it closes and is not rewritten afterwards, so a day you already hold stays valid.

Limits: Today's session as it trades is LIVE, not part of the files.

Older days in the levels API#

Why: one account for recent levels in a chart and the long history in a study. Where: the levels API's history call, with your account's key. Plan: The DATASET licence together with LIVE's levels API.

Ask the levels API for a day's history as usual.
Days older than the API's usual window answer when the account holds the licence.

What it measures: The levels API answers the last 90 calendar days of history for an account key; a day further back needs the DATASET licence.

Limits: The free archive on the site is untouched by this rule: every finished session stays free to read there.

Developer reference: /docs/developers/live

The licence

Dataset licence#

Why: the whole archive, all three gamma books, on your own machine, with no credits and no rate limit. Where: the licence on pricing; every field in the data dictionary. Plan: The dataset licence.

Start the licence from pricing; delivery starts that evening.
Download with the token you already hold for LAB, under your licence.
Read each field's meaning, unit, book and start date in the data dictionary.
Pick the book by its suffix: the volume, measured and open-interest books are never mixed.

The data dictionary: every field of the archive files and every LAB column, grouped by the volume, measured and open-interest books

What it measures: Every finished SPX session minute by minute, full per-strike detail in all three books, plus the market calendar with what usually followed each kind of event. A session lands minutes after the evening build.

How to read it:

Magnitudes are never comparable across books: a field's suffix names its book.
A field's start date is in the dictionary; some exist only from the day they were added.

Limits: For your own research. Showing the numbers to other people is licensing. Licensed raw vendor data is never part of it.

Developer reference: /docs/developers/dataset

Licensing for your product#

Why: the numbers inside your app, site, bot or paid channel, on published terms with no sales call. Where: Licensing. Plan: A redistribution licence, priced on the page.

Read the tiers on Licensing: levels only, the stream fifteen minutes late, or the stream in real time to an audience ceiling.
Press Request this licence; a countersigned email and a key follow.

The four licence tiers on /licensing: LEVELS, STREAM 15 MIN, STREAM and STREAM 250, each with its price and its ceiling of people

Tier	What it carries
LEVELS	Live levels each minute, ES and NQ, the archive as files; any audience
STREAM, 15 MIN	The book every second and the strike ladder with vanna and charm, fifteen minutes late; any audience
STREAM	The same in real time, up to 50 people
STREAM 250	The same in real time, up to 250 people
Larger audiences	Printed on the same page, up to 1,000 people and beyond

How to read it:

Attribution goes wherever the numbers are shown, with a link to firmtape.com; only the tier above 1,000 people drops it.

Limits: No tier covers a dealer-positioning terminal built on the stream, and none carries the raw vendor tape.

Add to DATASET

Add-on	What it adds
Redistribution licences	Show FirmTape's levels or its stream to other people in your app, site, bot or channel, on printed terms

Use it with other tools

Tool	How it pairs
REPLAY: The session archive and session pages	Read a session on the site before you study it in bulk.
LAB: The LAB vocabulary	Rebuild a LAB column from the fields it is built from, listed on the dictionary.
LIVE: Levels API	Recent history over the levels API; older days with the licence.

Good to know

The licence is for your own research: internal use, no redistribution, no competing service.
Delivery uses your account's token as a bearer header, from a script, with no browser.
Fields added later carry their start date; a study over all sessions should check it.

If something looks wrong

You see	Do this
A download is refused	The licence must be active on the account the token belongs to; check Account, Plans and billing.
A field is missing on an older day	It starts later: read its since column on the dictionary.
Two books disagree on the regime	They measure different things: the measured book disagrees with the convention book in a large share of minutes by design.

Glossary

Term	Meaning
volume book	cp-signed cumulative session volume per strike: the conventional public gamma object, built from the tape.
open-interest book	The same construction on the morning's open interest.
measured book	The book signed print by print from the tape and its quotes.
day file	One finished session as the terminal reads it, every minute and every strike.
data dictionary	Every field of the files and every LAB column, by book, with unit and start date.

Every reading here is a measurement of dealer positioning, never a forecast or a trade call.