Key insights
- Bitcoin transfer data can change a lot when analysts deal with change outputs, in ways.
- Public blockchain records do not clearly show why each movement happens.
- Researchers suggest using ranges of depending on one exact transfer estimate.
BIS researchers found that Bitcoin onchain transfer estimates can differ by up to sixfold. The September 15 study says measurement choices can materially change how analysts interpret blockchain activity. The findings also expose similar problems across Ethereum, Tron and stablecoin markets.
The research appeared in Working Paper 1377, titled Hidden by complexity? Measuring stablecoin, crypto and decentralized finance ecosystems. Timothy Aerts, Ronald Heijmans, Jan Paulick and Violeta Vuletic conducted the study using data from Bitcoin, Ethereum and Tron.
Bitcoin data changes with measurement methods
The researchers examined Bitcoin data from 2009 through 2026. Their dataset covered about 1.3 billion transactions and 3.6 billion transaction outputs.
Bitcoin’s UTXO structure creates a major measurement problem. Transactions often return unused funds to the sender as change. Therefore, counting every output can exaggerate the value transferred between economic participants.
The researchers tested three measurement approaches. The broadest method counted all transaction outputs. An adjusted method removed outputs sent back to the sending address. A more conservative method removed identified self-transfers. When researchers could not identify them, they removed the largest output.
Those approaches produced monthly transfer estimates that differed by as much as sixfold. However, the researchers said the lower estimate remains a heuristic rather than a definitive measure. They also identified CoinJoin transactions, mixers, spam and intermediary transfers as further complications.
Bitcoin valuation also produces major differences
The BIS study also examined Bitcoin’s market capitalization. Conventional market capitalization applies the current price to the entire circulating supply.
Realized capitalization takes another approach. It values each unspent output according to the price recorded when it last moved.
The difference can become substantial during major market moves. The study found conventional market capitalization reached as much as four times realized capitalization during rapid price increases.
The researchers also looked at dormant Bitcoin. Than 1.8 million Bitcoin stayed untouched for over fifteen years yet the study found that some of those coins moved after remaining dormant for more, than a decade.
This evidence point makes it hard to accurately determine the age of permanently lost bitcoins. Moreover, the researchers have noticed that dormant coins can still be of economic value without being used.
In the Bitcoin’s bear market of 2022, realized capitalization even surpassed the traditional market capitalization. That movement showed how different valuation methods can produce contrasting readings of market conditions.
Ethereum and USDT complicate blockchain analysis
Ethereum creates a different measurement problem through smart contracts. A single transaction can trigger multiple operations across contracts, token transfers and event logs. The researchers examined about 67.5 million active deployed contracts. More than 54 million remained outside the study’s technical classifications.
Among the contracts researchers classified, proxies and token contracts formed significant groups. The study also found extensive token naming duplication.
For instance, thousands of Ethereum contracts had USDT as their symbol. Only one was the Tether’s official issuance contract for Ethereum. Thus, single token symbol searches can yield an incorrect activity estimate.
USDT is also applied to different networks for other uses.Ethereum had a higher percentage of USDT that was locked up in smart contracts. That share exceeded 20% in 2022 before declining in later periods.
Tron showed a different pattern. Smart contracts held around 1% of USDT during much of the period examined. The researchers linked Ethereum activity more closely with DeFi functions.
Meanwhile, Tron activity appeared more consistent with transactions and store-of-value use. The researchers cautioned that account types cannot perfectly identify economic purpose.
BIS urges analysts to rethink crypto metrics
The study places its findings within a broader problem affecting blockchain analytics. Public records provide enormous amounts of information, but technical records do not always equal economic activity.
The researchers therefore recommend bounded estimates with clearly stated assumptions. They also call for technical classifications and separate analysis of assets and blockchain infrastructure.
Onchain indicators as explained by the BIS are not measures of economic activity but instead are noisy approximations. It is not a new regulation or a new reporting requirement, but rather its working paper.
Visa’s Onchain Analytics dashboard provides a comparable example. Its methodology separates total stablecoin activity from adjusted activity.
Visa says its adjusted figures remove potential distortions from bots, high-frequency trading, bridge routing and internal exchange operations. The dashboard also categorizes activity across payments, DeFi, exchanges, trading and other uses.
A September 16 snapshot cited in the supplied research showed a wide gap between total and adjusted stablecoin volumes. That contrast reinforces the study’s central finding that methodology can materially change headline metrics.
The BIS research therefore shifts attention from raw transaction counts toward the assumptions behind those figures. For analysts, the distinction matters because identical blockchain records can support very different estimates.
Conclusion
The BIS study shows that blockchain transparency does not guarantee simple economic interpretation. Bitcoin transfer estimates can vary sixfold, while market capitalization can diverge sharply from realized capitalization.
Ethereum and stablecoins add further complexity through smart contracts, duplicated token identities and different cross-chain use cases. Therefore, analysts should exercise more caution in interpretation of crypto activity and use a wider range and more clear methodology.
