NodeWitness Team

Every NodeWitness article carries the same byline, "NodeWitness Team." It stands for a small, independent project based in Spain that studies Bitcoin through on-chain data. We leave personal names off on purpose. A claim here should stand on the data it was checked against, and that data is something you can look up yourself. About the project goes into the reasoning.

Where the numbers come from

Most of what we publish is computed on a Bitcoin node we run ourselves, a full Bitcoin Core node with no pruning. That gives us the entire set of unspent coins (the UTXO set) without renting the core of our data from a third-party API. The node's current block height is shown on the network and mining page, so you can compare it with any public block explorer.

Metrics built on coin age or cost basis, such as realized price and HODL waves, come from a weekly snapshot of the UTXO set. Each figure is published with its date, and with the snapshot's block height where that applies. When a piece mixes readings from different weeks, it says so.

How a figure gets checked

• Fresh reads before publishing. Numbers are pulled again from the source or from our own API right before a piece goes out, rather than carried over from a draft or an earlier article.
• Incomplete snapshots are flagged. If more than 10% of a snapshot's supply can't be evaluated because a creation date or a historical price is missing, we mark it as degraded. We then either show the warning next to the figure or stay with the last complete snapshot and its date.
• Cross-checks against third parties. The MVRV we compute from our node, for instance, tracks Coin Metrics' version with a median deviation below 0.2% across the weekly series. When two sources disagree, we show both and explain the gap.
• Our data and outside data stay apart. Node figures are presented as ours and verifiable. Anything from elsewhere is attributed in the text and linked.
• Gaps stay gaps. A missing data point is never replaced with an estimate or a neutral filler value.

Sources

Any article that relies on outside data ends with a numbered source list, and each external figure is tagged [n] in the text. Entries are labeled primary (an official body, a manufacturer's technical advisory, a project's own repository) or secondary (press, aggregators). A secondary source never carries a conclusion on its own, and we don't link anything we haven't read.

Besides the node, the outside sources you'll see most often are Coin Metrics Community (part of the Score's market metrics), alternative.me (the Fear & Greed Index) and mempool.space (fees), plus Farside Investors for ETF flows. On security and protocol stories we go to the original document first, whether that's the vendor's advisory, the Bitcoin development mailing list or the code on GitHub. Price is market data from outside the chain, and we treat it that way.

Methodology and open data

• How the Score works lists its ten indicators, where each one comes from and how often it updates. The exact weights are withheld, for reasons set out on About the project.
• Score version history dates every methodology change.
• Where the Score has failed covers the cycle tops and bottoms it missed.
• On-chain charts, each with its method, its limits and a ready-made citation line.
• Technical and forensic research, where we trace real coins across the chain using a published method.
• Press kit with this week's figures, how to cite them and the license for each.
• On-chain glossary.
• The API behind the site is public and needs no sign-up. Node status lives at https://nodewitness.com/api/node/status, issuance at https://nodewitness.com/api/node/issuance and HODL waves at https://nodewitness.com/api/market/hodl-waves.

Corrections

When something we published turns out to be wrong, we fix it in the article itself and add a visible, dated note saying what changed. Errors aren't quietly edited away. If the same mistake shows up in several articles, each one gets its own note.

Recent examples:

• Vitalik and the cost of Bitcoin mining, October 27, 2026. Our node overstated the hashrate at the October 16 retarget, about 1,006 EH/s rather than 1,065 EH/s. The text, the chart and the hashprice figures were corrected.
• Coldcard's entropy flaw, October 22, 2026. The affected firmware range for the Mk3 and Mk2 v4 is 4.0.1 through 4.1.9, narrower than we first reported.
• Bitcoin's realized price, October 8, 2026. An earlier version implied realized price feeds the Score. It doesn't.
• Buying bitcoin in Spain for the first time, October 8, 2026. We called 280 satoshis a fraction of a euro cent, when it is closer to 20 cents.
• Where the Score has failed, October 6, 2026. Our price series was off by one day. We fixed it and refroze the backtest, and none of the conclusions changed.
• BIP-110's fork and replay risk, September 5, 2026. Replay protection on the split chain exists, but it is opt-in, not automatic.

Contact

Spotted a number that doesn't add up, or a claim that doesn't hold? Write to contacto@nodewitness.com. Nothing helps us more.