All news is rigorously fact-checked and reviewed by leading blockchain experts and seasoned industry insiders. - BIS researchers found large differences between methods used to estimate Bitcoin’s economic transaction value.
- Technical transfers, wallet change and internal movements can inflate raw blockchain activity.
- Ethereum and Tron show why even stablecoin statistics cannot always be compared directly across networks.
- For investors, the research makes methodology behind an on-chain metric almost as important as the number itself.
A Bitcoin transaction worth $10 million on-chain does not necessarily mean $10 million changed hands. That distinction sits at the center of new research from the Bank for International Settlements, which argues that crypto’s unusually rich public data can create an illusion of statistical precision.
In Working Paper 1377, published September 15, Timothy Aerts, Ronald Heijmans, Jan Paulick and Violeta Vuletic examine Bitcoin, Ethereum and Tron to determine how closely blockchain records correspond to underlying economic activity. Their conclusion is uncomfortable for anyone comparing networks using headline transaction volumes: the result can depend heavily on what the analyst decides to count.
The researchers identify three separate measurement problems: transaction aggregation, smart-contract programmability and cross-chain comparability. Their broader point is not that on-chain data is unreliable, but that raw protocol activity and economic activity are different datasets hiding inside the same ledger.
Why Sending 1 BTC Can Look Like Moving 10 BTC
Bitcoin’s UTXO architecture provides the clearest example.
Suppose a wallet controls an unspent output worth 10 BTC and wants to pay someone 1 BTC. It cannot simply remove one bitcoin from that existing output. The old UTXO is spent and new outputs are created. One could send 1 BTC to the recipient while another returns roughly 9 BTC to an address controlled by the sender, ignoring fees for simplicity.
An unadjusted calculation could therefore observe outputs approaching 10 BTC even though only 1 BTC represented the intended payment.
The problem becomes larger when exchanges consolidate wallets, batch withdrawals or reorganize funds between addresses they control.
The BIS finds that different approaches to these problems can produce Bitcoin transaction-value measurements varying by as much as a factor of six.
This is not merely theoretical. Glassnode documented a striking historical example in 2020: Bitcoin appeared to record more than $250 billion in daily on-chain volume on August 21, 2019, but the spike came from an exchange reshuffling funds and creating new cold wallets rather than an equivalent amount changing ownership.
That provides an independent, practical example of exactly the measurement problem the BIS is now formalizing.
What Traders Should Actually Check Before Using On-Chain Volume
For market analysis, the BIS paper changes how transaction-volume metrics should be read. A chart labelled simply “Bitcoin transaction volume” is not enough.
Before using the number as evidence of rising network demand, investors should check:
- Whether change outputs have been removed, particularly for UTXO networks such as Bitcoin.
- Whether transfers between addresses belonging to the same entity are excluded.
- Whether exchange wallet reorganizations and other internal movements remain in the dataset.
- Whether the methodology can revise historical observations as address-clustering models improve.
This is where commercial analytics providers already differ from raw blockchain statistics.
Glassnode’s entity-adjusted volume, for example, clusters addresses estimated to belong to the same network entity and excludes transfers within those clusters. The company explicitly warns that these metrics are based on heuristics and statistical techniques, meaning recent observations can change as its clustering improves.
Coin Metrics illustrates the other side of the problem. Its documentation for certain transfer-value metrics explicitly states that change outputs are not adjusted, meaning transfers back to the sender can remain part of the measurement. That does not make the metric wrong. It means analysts need to know what question that particular metric answers before comparing it with another provider’s data.
That is the practical information gain from the BIS research: two reputable dashboards can display different numbers without either dataset necessarily containing an error.
Ethereum’s 13 Million Contracts Create a Different Kind of Noise
Bitcoin requires analysts to decide what constitutes a genuine transfer. Ethereum forces them to decide what constitutes economically meaningful software.
The BIS researchers classified 13 million active smart contracts, including about 1.4 million tokens. Yet trading activity was highly concentrated and centered around stablecoins.
The gap between contract proliferation and concentrated activity matters when evaluating blockchain adoption.
A rising contract count could reflect genuinely new applications, but it can also capture token factories, duplicated contracts, abandoned deployments and technical infrastructure. Transaction counts face a related problem because one economically meaningful action can generate several technical interactions.
For investors comparing Layer 1 networks, “more transactions” and “more contracts” should therefore not automatically be translated into greater economic usage.
The useful question is what type of activity generated them.
Stablecoin Volume Needs a Chain-Level Context
The Ethereum-Tron comparison makes that point particularly clear.
The researchers found that stablecoins on Ethereum are more closely connected with smart-contract interactions.
On Tron, they are more commonly held outside smart contracts, which the authors say is consistent with transactional and store-of-value motives.
That means even the same stablecoin can represent different economic behavior depending on where it circulates.
A simple league table ranking blockchains by stablecoin supply or transfer volume can miss that distinction. For Ethereum, part of the activity may represent collateral movement, decentralized exchange trading or another application interaction. On Tron, a larger portion may reflect wallet-to-wallet transfers or dollar-like savings behavior.
The numbers can be comparable technically while being less comparable economically.
The Real Risk Is False Precision
The BIS paper ultimately matters because on-chain statistics are moving beyond crypto dashboards.
Stablecoin adoption, blockchain payment activity and DeFi exposure increasingly feed into research used by financial institutions and policymakers. If a methodology counts internal wallet movements as economic transfers or compares fundamentally different stablecoin use cases as equivalent activity, the distortion can migrate from a chart into an investment thesis or policy assessment.
For traders, the practical response is not to abandon on-chain metrics. It is to stop treating them as accounting statements.
Glassnode’s entity-adjusted approach demonstrates one solution, while Coin Metrics’ documentation shows why metric definitions need to be checked before datasets are compared. The BIS proposes the same broader discipline: explicit assumptions, technical classification and disaggregation rather than simple blockchain-wide totals.
That shifts the analytical question from “How much value moved?” to something more useful:
How much value moved between economically distinct participants, and what assumptions were required to arrive at that number?
For Bitcoin, the difference between those two questions can be severalfold.
