BIS Paper Flags Large Mismatch in Bitcoin On-Chain Transfer Data

New research from the Bank for International Settlements (BIS) suggests many of the headline metrics used to describe crypto activity—especially onchain “transfer” values—can be misleading depending on how the underlying blockchain data is counted. The BIS team reports that estimates of Bitcoin transfer values can differ by as much as six times when measurement methods change, driven largely by how transaction outputs are interpreted.

The study also highlights broader problems across the crypto ecosystem, extending beyond Bitcoin to Ethereum and stablecoins. BIS researchers warn that onchain indicators should often be treated as “noisy approximations rather than direct measures of economic activity,” rather than precision readouts of real-world flows.

Key takeaways

• Bitcoin onchain transfer values can swing by up to 6x based on how outputs—such as change back to the sender—are counted.
• Common market-cap style measures may overstate realized value; BIS finds conventional capitalization has at times been up to 4x higher.
• Ethereum’s smart-contract environment complicates classification, with tens of millions of active contracts that BIS could not categorize using the study’s framework.
• Stablecoin activity varies by chain and purpose, so aggregating across networks can blur how USDT is actually used.
• Some analytics providers already adjust raw volumes to remove distortions tied to behaviors like internal exchange routing or bot-driven activity.

Why “transfer value” can mean very different things

In the BIS working paper, the researchers focus on a measurement gap: when analysts try to estimate how much Bitcoin is being transferred onchain, the result depends heavily on the rules used to parse transactions. BIS’s key point is not that onchain data is absent, but that the same data can produce drastically different “economic activity” estimates.

The sixfold discrepancy reported by BIS is tied to differences in transaction measurement methods. One major driver is Bitcoin’s transaction structure. When a user spends Bitcoin, the transaction often includes unspent funds returned to the sender as a “change” output. Depending on the methodology, that change can be counted as an additional output—despite not representing value sent to another party.

BIS argues that this kind of counting convention can create the appearance of greater transfers than what actually reflects third-party movement. The researchers underline that metrics frequently presented as straightforward—such as transaction volumes, market capitalization, and total value locked—may carry more certainty than the structure of the underlying data actually supports.

Bitcoin market capitalization: a similar measurement mismatch

The BIS paper extends the measurement theme beyond transfer values to capitalization. The researchers report that a conventional market-cap approach has, at times, been as much as four times higher than realized capitalization.

According to BIS, realized capitalization values each coin at the price at the time it last moved. That distinction matters because it ties the valuation method to activity timestamps, rather than assuming a single uniform pricing snapshot. The implication for investors and market observers is that onchain-linked metrics can diverge from how value is actually being reflected in usage—especially when measurement assumptions are treated as neutral.

Cross-chain complications: Ethereum classification and stablecoin aggregation

While Bitcoin’s transaction design creates ambiguity around change outputs, Ethereum presents a different kind of complexity: smart contracts. BIS examined roughly 67.5 million active contracts and found that about 54 million could not be categorized using the classifications used in the study.

This matters because any attempt to interpret stablecoin flows or onchain transfers often depends on understanding whether activity belongs to known contract patterns—such as decentralized finance interactions, custody, payment services, or other use cases. When classification fails at scale, the risk increases that analytics will treat diverse behaviors as if they were homogeneous.

Stablecoins add another layer. The BIS researchers note that the same asset can serve different functions across networks. In their observations, USDT on Ethereum was more closely tied to DeFi activity, while USDT on Tron showed stronger association with payment-like and store-of-value purposes. BIS further highlights that the split is visible in smart contract holdings: in 2022, the share of USDT held by smart contracts on Ethereum exceeded 20%, compared with around 1% on Tron.

The practical takeaway is that aggregating stablecoin activity across chains can conflate distinct economic behaviors. BIS frames the resulting indicators as approximations that may obscure how stablecoins are being used in practice.

Overall, BIS’s conclusion is that onchain indicators should be approached as noisy estimates rather than direct measurements of economic activity—particularly when the indicators are presented as if they map cleanly to real-world transfers.

Adjusted analytics: how some dashboards try to correct distortions

Not all analytics treat raw blockchain activity as a final truth. Some providers attempt to separate “raw” transaction counts from adjusted volumes designed to better represent underlying economic activity.

Visa’s Onchain Analytics dashboard—powered by data from Allium Labs—shows both total and adjusted stablecoin transaction volumes. The dashboard’s adjusted methodology is intended to remove distortions from activity that may not reflect broad economic transfer, including high-frequency trading, bots, bridge routing, and internal exchange operations.

On the dashboard, Visa reports $6.4 trillion in total stablecoin transaction volume across the networks it tracks over the past 30 days, versus $313.1 billion in adjusted volume. The size of that gap illustrates the central theme of the BIS study: depending on counting rules and filtering approaches, “activity” can look dramatically larger or smaller.

Importantly, this does not automatically validate any specific methodology as “correct.” Instead, it reinforces BIS’s broader warning: without careful definitions and adjustments, common onchain metrics can overstate what the data actually means for economic interpretation.

For readers tracking crypto adoption using onchain indicators, the key next step is to pay closer attention to methodology—especially whether metrics account for change outputs, smart-contract classification limits, chain-specific usage patterns, and filtering for bot-driven or internal operations. BIS’s findings suggest that as dashboards and analytics products mature, the real differentiator will be how transparently they define what they measure and how their measurement choices shape the numbers.