How the standard works

The Subnet Multiple is a single ratio: market cap ÷ annualized real revenue. It answers one question — how many dollars the market pays for each dollar of disclosed external revenue a subnet earns.

A low multiple means a subnet is priced like a business. A high one means the price runs ahead of what the subnet actually earns today. When a subnet has no disclosed external revenue, the multiple is undefined — the cap is a bet on the future, and we say so plainly rather than inventing a denominator.

Market caps are pulled live from the Taostats dTAO pool API. Caps are denominated in USD using the live TAO price at fetch time. The “Caps updated” stamp on the home ledger reflects the most recent pull.

Because caps are live and each revenue figure is fixed at its disclosure date, the multiples move daily even when a subnet’s underlying revenue figure hasn’t changed.

Revenue means external customer revenue only — real dollars paid by users outside the network for a service. TAO emissions are excluded. Emissions are the protocol paying its own participants; counting them as revenue would let any subnet manufacture a low multiple out of thin air.

Revenue figures are curated and updated from team disclosures and third-party verification. Each ledger row carries its own “as of” date and a source link — expand any row to see them. Figures are fixed at their disclosure date until a newer one lands.

• TI ✓ Verified — confirmed by the TAO Institute, often corroborated by on-chain activity (e.g. buybacks). Gold is reserved for this tier only.
• self — team-disclosed and unaudited. Credible, but reported by the subnet itself.
• none — no disclosed external revenue. Multiple is undefined; the subnet is priced on belief.

Where a subnet discloses a partial period, we annualize it and show the basis on the row (e.g. “90d × 4”, “Apr 2026 × 12”, “MRR × 12”). Annualization assumes the disclosed run-rate holds for a year — a simplification, and one reason a high-confidence figure can still move.

The daily briefing opens with one word for the state of the network: RISK-ON, SELECTIVE or DEFENSIVE. It is a rule, not a view. Five signals are measured each day — breadth, net flow, momentum, concentration, and whether revenue-backed subnets are drawing flow faster than the rest. Each scores +1, 0 or −1.

A total of +2 or more is RISK-ON, −2 or less is DEFENSIVE, anything between is SELECTIVE. Confidence reflects how many of the five agree with the call, not how extreme any single reading is — a regime backed by four aligned signals is more trustworthy than one driven by a single outlier.

Every regime on the site expands to show the five measurements and their verdicts, so the call can be checked rather than taken on faith. The regime is stored with each daily snapshot, which is what makes it trackable over time.

SubnetAtlas stores one network snapshot per day and keeps the raw measurements, not the conclusions. Every archived briefing at /daily/<date> is rebuilt from that day's stored numbers using the same rules as today's — an old briefing is never recalculated with new data, and never back-filled for days we did not capture.

“What changed” compares two of our own snapshots. It is not derived from your previous visit, so every reader sees the same changes, and the first visit shows real history rather than an empty baseline.

Every revenue figure carries its age, and the age is shown wherever the figure is. Age is measured from the end of the period the revenue covers, not from the date it was published — a figure released last week about March's revenue still describes a business six months ago, and the business is what we are pricing.

• Current — under 30 days old.
• Aging — 30 to 90 days old.
• Stale — over 90 days old.

Stale does not mean wrong. It means the multiple is comparing a live market cap against an older disclosure, so movement in that multiple is the cap moving, not the business being re-measured. Where we can attribute a change, we label the driver — CAP when only the valuation moved, CAP + REVENUE when a new figure landed too.

Expand any ledger row or subnet scorecard to see the source, the confidence tier, the period the revenue was earned in, when that period ended, the annualization applied, and a link to the disclosure. We record the publication date separately from the period and leave it blank where it was never captured rather than inferring it.

Subnet pages show a scorecard of separate measurements, not a single 0–100 rating. Collapsing a revenue multiple, pool depth, capital flow and momentum into one number would require us to assert how much each matters relative to the others, which we cannot do honestly on your behalf. The scorecard states each measurement and leaves the weighing to you.

Descriptive bands such as “priced like a business” and “belief premium” are labels over ranges of the multiple, defined above. They are not ratings and not recommendations.

The two halves of the ratio do not update at the same speed, and pretending otherwise would be the easiest way to mislead with this number. Caps are live. Revenue is not.

Every revenue figure currently on the ledger dates from Sep 2026 – Oct 2026. Those denominators stay frozen at their disclosure date until a newer disclosure lands, so a multiple can move a long way on the cap alone while the business behind it has not been re-measured in months. The ledger stamps this on the masthead, each row carries its own “as of” date, and the 7-day delta on each row names which side actually moved — cap or revenue.

• Emissions as revenue. TAO emitted to a subnet's own participants is the protocol paying itself. Counting it would let any subnet manufacture a low multiple.
• Buy or sell calls. There is no rating, no target, no signal to act on. The multiple is an input to your own work.
• Flow windows we cannot source. Where flow appears on this site it means one specific thing: net TAO into a subnet's pool over 24 hours, computed as buy volume minus sell volume, emissions excluded. It is bounded by the pool it moved through and by the volume that produced it. Taostats also exposes 1-hour, 7-day and 30-day “net flow” fields; we used to render them and they were wrong by orders of magnitude — one field alone implied more TAO moving through a single subnet in a day than exists in circulation. We no longer publish any flow window we cannot bound and verify.
• Revenue we scraped or estimated. Figures come from disclosures and third-party verification. If a subnet has not disclosed, the multiple is undefined and stays that way.
• Flow as a raw percentage of pool. The site used to print lines like “strong 24h inflow relative to pool size (+1,688,349.3%)”. The numerator was wrong, but so was the idea: a flow measured against a near-empty pool produces an enormous number that means nothing. Flow scale is now reported in bands — moderate, strong, exceptional — and pools too thin for the comparison to carry information are excluded from it rather than divided by one.

The home ledger shows the tracked dataset, not the full network. The thesis stat is derived live: the 5 entries with a disclosed revenue figure, against the live subnet count from Taostats. Entries with an announced deal but no figure (and subnets we don’t yet track) count as priced on belief. The full directory lives at /subnets.

Multiples are a snapshot, not investment advice. Figures change as disclosures and caps update. Verify against primary sources before acting.