How we verify holdings (our data methodology)

Treasury data gets cited, screenshotted, and quoted by journalists and analysts. A wrong number published under our name is worse than no number at all. So we hold ourselves to one rule: every holding and every move we publish traces back to an authoritative primary source — never to an unverified rumor, and never to another aggregator quoting an aggregator.

Different kinds of holders disclose in different places. We go to the most authoritative source available for each:

• Public companies →their filings with the US Securities and Exchange Commission, plus the company’s own investor-relations or treasury page. Filings are dated, exact, and legally binding.
• Exchange-traded funds →the fund sponsor’s own daily holdings reports and independent flow trackers, cross-checked against each other.
• Crypto-native private companies →on-chain records where a wallet is publicly known, and the company’s own disclosures. Where only an industry tracker has a figure, we treat it as a headline reference and label it as such — never as a precise, dated transaction.
• Governments → official central-bank and treasury disclosures.

We don’t just record the latest headline figure. For every company and asset, the individual buys and sells we’ve logged should add up to the current holding:

A check runs every day comparing what we’ve recorded against an independent reference. If the two disagree by more than a small tolerance, the case is flagged for a human to investigate before anything reaches the public site or the data feed. The goal is that no figure goes out the door unless the math behind it ties out.

We also use the date the move was officially filed, not a guessed transaction date, so a purchase lands on the timeline exactly when the public record says it did.

Not every source is equal, so each move carries a trust level, and the surfaces you read are filtered accordingly:

• Primary, dated, and exact — a specific filing, a sponsor report, or an on-chain transaction. These power the alerts, the activity feed, and the newsletter, because we can stand behind each one as a discrete, verifiable event.
• Headline reference figures— an industry tracker’s total for a private holder with no public per-transaction record. We use these to keep a leaderboard total honest, but we deliberately keep them out of the per-event feeds so a rough aggregate never gets mistaken for a confirmed purchase.

Honesty about limits is part of trust. A few things we deliberately do not do:

• We don’t invent a precise transaction from a round headline number. If we only have an approximate total, we say so.
• We don’t backdate a market-sentiment reading. When we show a fear-and-greed context for a historical purchase, we use the value from that date — or show nothing if no reliable history exists — rather than pasting today’s reading onto a past event.
• We don’t treat the slow, continuous coin growth from staking as if it were a purchase event.

You don’t have to take our word for it. Every move on the site carries a “View source” link that opens the underlying filing, sponsor report, or transaction. You can follow any number we publish back to its origin.

And you don’t have to trust that a number stays right after we publish it. Every change to a holding — the figure before, the figure after, and the moment it changed — is posted to a public audit log as it happens. When we get something wrong and fix it, the correction goes on a permanent accuracy ledger with the source that settles it — because a tracker that never shows its mistakes is hiding them, not avoiding them.

For the full picture of which sources we use, how often each refreshes, and the known limits of the dataset, see our data coverage & methodology page, and our trust & security overview for how we handle the rest.