This page is documentation, not a storefront: how the catalog is graded, how auth and access modes work, every endpoint and error code, and the full list of metals/macro data feeds the API serves. We sell data, not trading signals — validated trading signals to date: 0 proven, 1 candidate (Monero mean-reversion) in paper-testing.
Pricing → Access & Pricing on the homepage.
Every data feed is graded by crowdedness at build time — that grade is fixed. Whether an individual feed is currently 🔒 locked is a separate, live flag the catalog pipeline sets as it grows; it skews heavily toward Alpha but isn’t exclusive to it. Access tiers built on this grading are priced on the homepage — see Access & Pricing.
Commodity-grade. Funds already own this data; it’s here for completeness and to build a baseline model.
Contested. Real edge, but more than a few desks already watch it.
Uncontested. Small-capacity, non-purchasable, or outside standard vendor coverage — the focus of this catalog. A crowdedness grade on the feed, not a claim of validated alpha.
Real per-category counts, computed live from the catalog. “Locked” is how many of that category’s feeds currently carry 🔒 — the rest is already free to browse.
A distinct box: 10 metals/macro feeds + 4 derived series, served over REST. 7 standard, 7 flagged premium 🔒. Each product supports one or both access modes below (range and/or live). Pricing for these products lives on the homepage — Access & Pricing.
Every product supports one or both modes. Pick the one that matches what you’re building: a backtest wants range, a live model wants live.
Buy a fixed historical window. Your entitlement is {mode:"range", start, end, call_cap}. Queries outside the purchased window return 403 outside_purchased_range.
/latest is not available — range keys are historical-only. Best for backtesting.
Rolling access. Your entitlement is {mode:"live", window_days, rpm, per_day}.
/latest returns current values; /v1/data/{product} returns the trailing window_days. Enforced limits: rpm (per-minute) and per_day (rolling 24h) — both return 429 once exceeded.
Auth via Authorization: Bearer <key> header, or ?api_key=<key>.
401 invalid_or_missing_key · 403 key_expired · 403 not_entitled · 403 outside_purchased_range · 403 latest_requires_live_mode · 429 rate_limited · 429 daily_cap_reached · 429 range_call_cap_reached · 404 unknown_productBase URL above is illustrative — the real endpoint and your key are issued when you request access below.
Two independent access modes, matched to two different jobs. Rangeis a one-time purchase of a fixed historical window — you own it, /latest is disabled on that key, and it’s built for backtesting: pull the years you need once, test against them as long as you want. Liveis the subscription — a rolling window plus real-time /latest, rate-limited by rpm and per_day, built for running a strategy that needs today’s value tomorrow morning.
Rule of thumb: buying your first dataset to test an idea → range. Wiring a signal into something that trades on a schedule → live. Nothing stops you from buying both on the same product.
This page is documentation only. All plans, packages, and per-product pricing live in one place — Access & Pricing on the homepage. That’s the single source of truth for what things cost.
Yes. The catalog itself is free to browse in full — every currently-unlocked signal (386 of 1,543 today), no account needed. On the data-API side, /v1/sample/{product}serves a 30-day-delayed sample of any single feed, no key, capped calls/day — enough to see the actual shape of the data (real columns, real cadence, real gaps if there are any) before you decide what to buy.
No — they are data feeds. Every feed in the catalog carries a mechanism and a crowdedness grade, but a catalogued feed is a dataset, not a validated trading signal — see /about for the full five-stage validation process.
As of today, zero signals have cleared that process — 0 proven, with 1 candidate (Monero mean-reversion) in paper-testing. A publication-date rerun of the full search produced zero signals above threshold; one lead-lag pairing of 27 passed re-verification, marginally (/leadlag); and one published signal was withdrawn after re-verification identified a scoring error (/signal-live). Candidates flagged as promising are provisional and forward-tracked in the Research Lab, not sold as proven.
None of this has traded with live capital — everything on this site is paper-tracked research, not a track record, and nothing here is investment advice.
Tell us which domains or data-API products you want and we’ll issue your key directly — access is provisioned by hand while self-serve checkout comes online. First access goes to the waitlist in order. Pricing for what you’re requesting is on the homepage.