A new working paper from the Federal Reserve Bank of Philadelphia reveals a stark divergence in how Bitcoin and Ethereum markets react to public notifications of large cryptocurrency transfers. Published this month, the study finds that non-whale Bitcoin wallets rapidly follow the trading direction of identified whales, while Ethereum market participation remains largely stable.
Study Methodology and Whale Definition
The Philadelphia Fed researchers matched timestamps from Whale Alert public notifications with on-chain transfer data for Bitcoin (BTC), Ethereum (ETH), and Wrapped Bitcoin (WBTC) through the end of 2025. The authors defined a whale wallet as an address that had executed at least one transfer valued above $50 million, explicitly excluding large wallets associated with centralized exchanges or smart contracts.
To isolate distinct events, the study filtered for transactions without another whale transfer occurring within a two-hour window on either side. This process yielded a final sample of 6,645 Bitcoin whale transactions and 5,075 Ethereum whale transactions.
Bitcoin: Sharp, Short-Lived Herding Behavior
The data shows that active participation from non-whale Bitcoin wallets—specifically small and medium-sized cohorts—surged most intensely during the first 15 minutes following a whale alert.
Buy-Side Reaction (Following Whale Buys)
- Small wallets: Buy participation increased by 14.81 percentage points.
- Medium wallets: Buy participation increased by 23.72 percentage points.
- Large wallets: Buy participation increased by 3.50 percentage points.
Sell-Side Reaction (Following Whale Sells)
- Small wallets: Sell participation rose by 12.95 percentage points.
- Medium wallets: Sell participation rose by 29.52 percentage points.
- Large wallets: Sell participation rose by 2.95 percentage points.
This same-direction trading activity decayed toward baseline levels within approximately one hour.
Ethereum: Muted and Stable Response
In contrast, Ethereum did not exhibit a broad-based retail reaction. Post-alert participation remained comparatively stable across all wallet size groups. The only statistically notable immediate response appeared among the largest non-whale cohort following whale sells. Medium-sized ETH sellers registered a reaction only at the study’s weaker 10% significance threshold.
The authors emphasize that these wallet classifications reflect transaction-based proxies for activity levels, not the verified identities of the individuals or entities controlling the addresses.
Diverging Volatility Dynamics
The market structure difference extends to realized volatility:
- Bitcoin: Whale alerts correlated with a temporary rise in realized BTC volatility at short horizons. However, by the 24-hour mark, the volatility effect from both BTC and ETH alerts had reversed.
- Wrapped Bitcoin (WBTC): Alerts for WBTC on Ethereum showed a volatility impact statistically indistinguishable from zero.
- Ethereum: Realized volatility on the Ethereum network was lower after alerts, suggesting large Ethereum-network transfers tend to occur during periods of declining volatility.
Market Structure, Not Consensus Mechanism
The authors attribute the behavioral gap to fundamental market-structure differences. They note that Ethereum activity frequently routes through exchanges, smart contracts, and Layer-2 scaling solutions, where numerous user transactions are often aggregated into larger balance transfers.
This structural contrast persisted through Ethereum’s September 2022 transition to proof-of-stake, indicating that the consensus mechanism alone does not explain the divergence in market dynamics.
Observational Evidence, Not Causal Proof
The researchers caution that the evidence remains observational. Key limitations include:
- Wallet-size groups serve as transaction-based proxies rather than definitive entity classifiers.
- A single owner may control multiple addresses.
- Exchange-related activity was excluded from the whale definition and analysis.
Consequently, the event study establishes robust patterns in wallet activity and volatility surrounding public alerts, but does not prove that the alerts caused every observed market response.
