A recent working paper from the Bank for International Settlements (BIS argues that many of the headline figures used to describe cryptocurrency and decentralized finance activity can mislead more than they illuminate.
Drawing on roughly 100 billion records from Bitcoin, Ethereum and Tron, the researchers conclude that widely cited indicators of transfer volume, market size and on-chain usage are highly sensitive to how raw blockchain data are cleaned and interpreted.
As a result, they say these numbers should be treated as noisy approximations rather than precise gauges of real economic activity.
The most striking illustration involves Bitcoin.
Because the network uses an unspent-transaction-output model, a typical payment often returns unused funds to the sender as change.
Whether that change is counted as economic transfer value, excluded, or treated under a conservative lower-bound rule produces monthly estimates that can differ by as much as a factor of six.
Unadjusted totals therefore risk inflating the impression of how much value is actually changing hands between distinct economic parties.
Conventional market-capitalization figures can also diverge sharply from realized capitalization, which values coins at the price of their last movement; at times the gap has reached a factor of four.
Ethereum presents a different set of problems.
Programmability allows a single transaction to trigger many contracts and emit numerous events, making it hard to separate genuine economic use from technical by-products.
Researchers classified about 13 million active contracts, including some 1.4 million tokens, yet tens of millions of additional contracts defied straightforward categorization.
Token tickers themselves are unreliable: thousands of contracts have used the “USDT” symbol without any connection to the well-known issuer.
Trading remains highly concentrated and heavily oriented around stablecoins, further complicating attempts to read raw volumes as a map of underlying demand.
Stablecoins illustrate the cross-chain comparability problem. The same token can play different roles depending on the network.
On Ethereum, a sizable share of USDT has sat inside smart contracts, consistent with use in lending, liquidity pools and other DeFi operations.
On Tron, holdings have more often remained outside contracts, pointing toward simpler transfer and store-of-value motives.
Aggregating activity across chains without accounting for these differences can therefore mix distinct economic functions and overstate, or mischaracterize, overall usage.
Earlier related work on Ethereum stablecoin event logs reached a similar conclusion: a large fraction of recorded transfers occur inside complex, multi-step transactions rather than as standalone payments.
Treating every transfer as an independent payment therefore overstates both counts and volumes.
The authors identify three structural sources of divergence: how transactions are aggregated on a given chain, the open-ended nature of smart contract code, and the difficulty of comparing use cases across networks.
Simple sums, they argue, frequently blend genuine economic activity with technical artefacts.
For researchers, market monitors and policymakers, the practical implication is caution. On-chain dashboards remain valuable, but only when accompanied by explicit methodological choices and an understanding of what those choices leave out.