Methodology, coverage and limits.

What a report is built from, what each number is a share of, and what no report can establish.

Supported networks

The holder map covers every holder the provider returns; a smaller set of wallets is investigated in depth, and the report says which and why.

Mapped: The holder’s balance and share of the supply are read at the snapshot. Nothing about its history is traced.
Investigated in depth: The wallet’s observed funding and its trades on supported venues are traced, within the scope and budget of the investigation type. When a budget or a provider limit stops the work early, the run ends partial and the report lists what was not reached.

A number without its denominator misleads. Three different totals appear in a report:

Share of supply is measured against the total supply at the snapshot.
Holder counts are measured against the holders the provider returned. If that list is incomplete, the report says so and treats the count as a lower bound.
Pattern counts, such as wallets that bought the same way, come only from the wallets investigated in depth. Wallets outside that set were not checked, so the true number can be higher.

Robinhood Chain: An address with a balance above zero at the snapshot. Where no holder index is available, the list is rebuilt from the token’s transfer history and checked against the contract’s own figures. An address can be a person, a contract, a pool or an exchange: one address is not one person.
Solana: An owner wallet. Token accounts for the mint are grouped by the wallet that owns them, and empty accounts are left out. As on any network, one wallet can stand for many people and one person can use many wallets.

Observed funding means inbound transfers inside the history window of the investigation. A wallet’s first-ever funding is shown only when the provider can read that far back; otherwise it is marked unavailable.
Tracing stops at labelled services and at senders that fund many wallets, such as exchanges. Those are weak signals, not connections.
Some funding is not visible at all, for example certain bridge deposits and internal transfers. The report lists these gaps.

Observed: A record that was read: a balance, a transfer, a trade, a domain registration date.
Possible connection: Records that point the same way. It is a hypothesis about a relationship, never a statement of common ownership.
Alternative explanation: The ordinary explanation that fits the same records, printed beside the finding it challenges.
Unable to verify: Something that could not be read or checked. It is not a finding for or against the token.

Listing links and website text are published by the project itself and are reported as claims, not facts.
Only a limited part of a website is read, and claims made in images, videos, chats or PDFs are not read.
X posts are a limited recent sample. Followers and engagement can be bought and count as weak signals.

Implemented: Rule-based scans on both networks, including anonymous Quick Scans; AI analysis and the Skeptic’s review in the Full AI investigation; saved private reports; redacted summaries an owner chooses to publish.
Opens soon: Credit purchases, and with them paid runs for the public. No payment provider is connected yet.
Planned, not built: The holder benefit, which starts when the Ocellus token launches, and cross-case wallet memory.

Who owns a wallet
Shared funding, timing or buying patterns can suggest a link. Ocellus never presents them as proof that one person controls two wallets.
Whether a token is a scam
A report lists observations and what could not be verified. It never labels a token, a team or a wallet a scam, and it gives no risk score.
That no flags means safe
A report with few observations means little stood out in the records that were read. It says nothing about whether a token is safe to buy.
That a check is always right
New checks are replayed on past cases before they ship, and they can still be wrong. Findings keep their strength labels and caveats.
Where the price goes next
No predictions and no trading signals. Reports are not investment advice.
What the records do not show
When a data provider, a budget or a scope limit stops a step, the report says what was not covered instead of filling the gap.