sats/share

Methodology

The metric

For an entity holding bitcoin directly, sats per share is BTC held × 100 000 000 ÷ common shares outstanding, both from that entity's own SEC filings. For a fund, it is the sum over every disclosed position with non-zero exposure of position shares × sats per share of the holding, divided by the fund's shares outstanding. The recursion resolves depth-first with cycle detection and a depth cap of four hops.

The metric is denominated in sats, so it does not move with the bitcoin price. It moves when a company buys or sells bitcoin, when shares are issued or retired, or when a fund rebalances — and each of those becomes knowable only when a filing lands.

Sources

Balance-sheet bitcoin: 10-K, 10-Q and 8-K filings, read and confirmed by a person before any number changes — extraction is proposed, reviewed, then published, never automatic. Fund holdings: Form N-PORT XML, which is structured and ingested automatically. Fund share counts: the fund's own audited annual reports. The BTC price (used only to express sats in dollars, never in the metric itself) comes from Kraken's public API; if that fetch is more than seven days old the site refuses to build rather than ship stale as current.

Vintage — the honest date

A look-through figure blends inputs of different ages: a company's holdings might be weeks old, a fund's positions up to five months (N-PORT becomes public sixty days after quarter-end, and only the third month of each fiscal quarter is published). Every figure here carries the date of its oldest contributing input, displayed beside it, not in a footnote. Where the inputs span more than 90 days the figure is additionally flagged. A fund's position sizes are never extrapolated between filings; a holding's own sats per share updates when the holding files, but the composite's vintage stays pinned to its oldest input.

Confidence

Mechanical, not judgement: High — direct holder, holdings filed under 45 days ago. Medium — one hop, all inputs under 120 days. Low — two or more hops, any input over 120 days, or an input spread over 90 days.

Filed vs inferred

A figure is filed when the unit count is literally printed in the cited document. It is inferred when derived (a stated sum of stated components, for instance) — and it is marked as inferred everywhere it appears, including exports. Where a filing states only dollar values and no unit count, the period is left as an honest gap rather than estimated.

Split adjustment

Historical series are adjusted to the present share basis using ratios from 8-K filings, in exact fractions. The adjustment factor is published beside every point, so the raw filed numbers reconstruct exactly: raw = adjusted ÷ factor. The test case is Strategy's 10-for-1 split of August 2024 — the adjusted series crosses it at ×1.07 (a real quarter) where the unadjusted series shows a ×0.11 cliff that never happened.

The universe

Coverage is derived from the filings, not curated: seeds (every US-listed balance-sheet holder and spot ETP) are found in funds' N-PORT filings via identifier joins — never name matching — then funds holding those funds join, to depth four. Excluded: non-US listings without SEC filings; derivative exposure (notional-to-exposure conversion is a modelling choice — excluded and disclosed per position); leveraged and inverse products from rankings.

OTC-quoted holders are excluded: this site covers US-listed entities, and OTC quotation is not an exchange listing. Four filers with bitcoin on the balance sheet are held out under this rule — CIMG Inc. (1 145.4 BTC), Athena Bitcoin Global (9 BTC), Northern Minerals & Exploration Ltd. (0.50096 BTC), T-REX Acquisition Corp. (0.0016 BTC). Their filings are linked so the exclusion is checkable, not silent.

Adaptations worth knowing about

Operated by Delta Gamma LLC. Nothing here is investment advice; verify against the linked filings before relying on any number.