Bring source-linked DePIN observations into your research process. Evaluate coverage, history, measurement definitions, provenance, integration requirements, and permitted use before committing to access. Each capability below identifies the question it addresses, the inputs it uses, and its limitations — a detector's existence does not establish every economic explanation attached to it.your data vendor should never be the weakest link in your risk model.
FDV/MC classification over 340 networks with market data (of our 370 live-data networks): which positions carry valuation-weighted supply overhang. Unlock dates require separate schedule evidence. Live distribution right now: 11 EXTREME (>90% unreleased — currently topped by LNQ at 2223194504181319936x), 20 HIGH, 67 ELEVATED, 80 LOW, 162 MINIMAL. Sector aggregate: $13136880780347456B FDV against $486B circulating — this ratio measures valuation-weighted overhang, not a fraction of token units across networks, and individual networks require separate supply and unlock-schedule checks.
The integrity gate: an upstream source reported a network FDV of $4.42 septillion — 18 orders of magnitude off. The engine now reconstructs FDV from price × fully-diluted supply (both measured first-party) and rejects any upstream value that disagrees by >10x, recording the rejected value, the reason, and the recomputation in an explicit fdv_correction block. A silent correction is the same failure as a silent error — you see every override we make.
The slow bleed pattern in DePIN tokens: utilisation falling while node count grows — oversaturation eroding node revenue before it shows in price. The detector runs over our first-party utilisation and node-count telemetry and flags the divergence. We publish it as the detector we run — with the query and the verify URLs — not as a constant live finding; on current AKT data the oscillation has no trend (we checked, and we say so on the record).
Upstream sources silently rewrite history. We keep the evidence: 39 restatement events, machine-detected by prior-value diff over published series (networks observed restating so far: 13, of our 370 with live data). Live case: Arweave’s cumulative block count was rewritten 5 times — a published date of 1,992,299 blocks replaced with 1,702,218 (−14.56%) for the same date; a cumulative counter that cannot legally go down. The node reports ~1.99M again today; only a before/after ledger proves the restate ever happened. If your backtests and risk models consume vendor data, this ledger is how you know what they were built on.
Every published value traces to raw upstream bytes: raw_sha256 over the fetched payload, code_sha256 over the collector build, replayable via the derivation ledger (3,708 pins / 248 symbols in the last 24h). The serving path is deterministic code — zero ML between upstream bytes and published values — which is exactly why a code hash can pin the entire transform. Re-derive any value yourself: GET /v1/derivation?symbol=AKT.
Node-quality grades (A–F, quality-adjusted counts), GPU supply/demand, sector health, geo concentration, dev velocity, competitive landscape — every engine, unrestricted, plus the full API, snapshot history, and the not_covered map.
What our instruments proved this month, stated at exact strength (all machine-verified, none hand-waved):
All five capabilities above, unrestricted: dilution map, divergence detector, revision ledger, provenance rail, full engine suite. Full API + snapshot history + direct line to the operator. Position-level data for a book, not a dashboard.
Pay now — instant checkout:
Pay $2,000/mo now Pay $20,000/yr (2 months free)
Card checkout opens instantly. Annual is $20,000 vs $24,000 month-by-month — the two-months-free math is exact. Prefer an invoice, PO, or wire? Contact the operator — same price, human paperwork.
We will not quote you a competitor’s price or a salary figure we cannot source, so here is the honest comparison: $2,000/mo is one seat’s worth of budget, priced against what a specialist data feed costs rather than against a headcount. We do not replace your data team. We replace one line item — collecting, normalising and verifying DePIN supply-side telemetry — every supported value carries its source, timestamp, and verification path, so researchers can inspect the evidence directly.
Evaluate first: Free verification tier — confirm the data against upstream verify URLs before paying, at every tier.
Lighter tiers exist and always will: Design Partner $199/mo (forever lock, 0/20 claimed), Pro $499/mo, agent-credit metering from $0.02/query. The institutional tier is for desks that need the whole engine suite and the ledger.
We do not sell predictions, signals, win rates, or alpha. We have no forward-tested basis for a performance claim, so we do not make one — our own claim checker refuses that entire class of statement at any number. We are not a broker-dealer, an RIA, or a research analyst, and nothing we serve is advice. This is machine-readable measurement and provenance. Stated as the shape of the offering, not as a gap.
Every value we serve carries source, as_of, and a verify_url pointing at the upstream endpoint you can call yourself. No estimates, no zero-fill, no back-fill: if a metric isn’t flowing, it’s absent — not zero. A restatement is kept as evidence, never overwritten. For a fund, that is the difference between a risk model built on data and a risk model built on a vendor’s current opinion of the data.
What we do not claim: directional alpha, win rates, or signal accuracy. No forward-tested basis for such claims exists on this estate, so none is published. We sell measurement, integrity, and risk mapping. The volatility-memory discovery is published as a finding about market structure — not as a trade signal.