The River report landed with a thud: 70% of Bitcoin is held in self-custody. The number was repeated across crypto Twitter, cited as evidence of a mature, decentralized market. It is a lie dressed in statistics. Not a deliberate lie, perhaps, but a methodological one. The truth is more uncomfortable. Self-custody is a spectrum, not a binary. And the report's definition conflates ownership with control, ignoring the cold reality of key management, inheritance planning, and the silent custodianship of exchanges. In my 2022 analysis of Terra's collapse, I learned that the gap between perceived and actual liquidity is where risk lives. The same gap exists here. Let me walk you through the math.
Bitcoin's circulating supply is 19.5 million. The River report claims 70%—13.65 million BTC—is in self-custody. That implies only 30% is on exchanges, in ETFs, or in other custodial arrangements. But consider: the eleven US spot ETFs hold over 1.1 million BTC. Major exchanges—Binance, Coinbase, Kraken, Bitfinex—collectively hold at least 3 million BTC in hot and cold wallets. MicroStrategy holds 214,000. The Mt. Gox estate still holds 142,000. These are custodial or corporate holdings, not self-custody in any meaningful sense. That's already 4.5 million BTC. Add the estimated 1.5 million BTC lost or inaccessible (Satoshi's coins, forgotten wallets, burned addresses), and we are at 6 million. The remaining 13.5 million is distributed across millions of individual addresses. But not all of those are self-custodied.
The core insight: The report likely counts any BTC not held on a known exchange as self-custody. This is a structural error. A wallet that has not moved in three years may belong to a dead person, a lost key, or a custodial service that does not publicly disclose its addresses. The UTXO age distribution tells a different story. According to CoinMetrics, as of Q1 2025, 35% of UTXOs have been dormant for over five years. Some of these are long-term holders, but many are lost. The report's methodology fails to account for this. During my 2023 Solana audit, I quantified centralization vectors by simulating transaction flows. The same approach applies here: simulate the probability that a given UTXO is truly self-custodied. The result is a confidence interval, not a single number. The River report's 70% implies a confidence that data does not support.
The contrarian angle: The report is not entirely wrong. The trend is real. The percentage of Bitcoin held in self-custody has increased since the FTX collapse. In 2022, it was likely around 50%. By 2025, it may be 60%. But the report's aggressive estimate ignores the growing institutional custody sector. The ETF approvals in 2024 brought billions into regulated custody, which is not self-custody. The report might have intended to highlight the resilience of the individual holder, but it overshot. The truth is more nuanced. Self-custody is growing, but so is institutional custody. The two are not zero-sum. The market is bifurcating.
Takeaway: The River report should be seen as a marketing document, not a scientific audit. The raw data and methodology remain unpublished. Until they are, treat the 70% figure as a hypothesis, not a fact. Certainty is a luxury; risk is the baseline. The real question is not how many Bitcoin are self-custodied, but how many can be moved without permission. That number is likely lower than we think. And that is the number that matters.