A Philadelphia Fed working paper found that non-whale Bitcoin wallets were more likely to trade in a whale's direction within 15 minutes of a public whale alert, with the same-direction reaction fading toward baseline inside an hour and barely registering for Ethereum. The finding, central to the Bitcoin whale trading debate, is drawn from a preliminary event study and describes cohort participation shares — not verified copy-trading by individual holders.

Key Points

• Philadelphia Fed Working Paper 26-42 reports that non-whale Bitcoin wallet cohorts were more likely to trade in a whale's direction shortly after a public alert.
• The strongest reaction lands in the first 15 minutes after a public whale alert and wanes toward normal within 60 minutes.
• The paper is preliminary and observational; the estimates are participation-share changes across wallet cohorts, not proof that traders copied whales.

What the Philadelphia Fed Paper Reports About Bitcoin Whale Trading

The finding appears in Philadelphia Fed Working Paper 26-42, titled "How Do Large, Sophisticated Cryptocurrency Trades Impact Broader DeFi Market Dynamics?" Authored by Keith Hazen, Julapa Jagtiani and Loretta J. Mester, it was published in September 2026n

The Reported Same-Direction Trading Pattern

Using the public dissemination of a whale alert as the event clock, the study measures whether non-whale wallets subsequently traded in the same direction. In the full-sample Bitcoin regressions, whale-buy alerts were associated with first-15-minute increases in non-whale buy participation of 14.81 percentage points for small wallets, 23.72 percentage points for medium wallets and 3.50 percentage points for large wallets, all significant at the 1 percent level, as CryptoSlate first reported on September 11, 2026.

The sell side mirrors that pattern. Following Bitcoin whale-sell alerts, the first-15-minute estimates were 12.95, 29.52 and 2.95 percentage points for small, medium and large wallets, significant at the 5 percent, 1 percent and 5 percent levels respectively. These are cohort participation-share changes, not individual probabilities or token returns.

The 15-Minute Window and Its Trigger

The trigger that starts the clock is the public dissemination of a Whale Alert notification, not a private data feed. Whale Alert is a blockchain-tracking service that publishes large on-chain transfers as they happen, so any observer sees the same signal at the same moment — the property that lets the study time reactions against a common clock. The raw Bitcoin directional response is strongest in that first 15-minute window and decays toward normal within 60 minutes — a short half-life consistent with reaction to a public signal rather than a durable structural shift.

Ethereum behaves differently. The paper describes Ethereum's immediate directional response as largely absent except among the largest investors in the sell model, where the estimate is 0.76 percentage points, and notes that the Bitcoin-Ethereum contrast persists across Ethereum's September 15, 2022 transition from proof of work to proof of stake. For readers inclined to treat whale alerts as a market-wide signal, the divergence is the relevant caveat: the alert-driven reaction documented here is largely a Bitcoin phenomenon in this sample, not a uniform crypto-market reflex.

What the Reported Pattern Can Tell Us About Wallet Behavior

The authors frame the result as a market-structure signal. In their abstract, Hazen, Jagtiani and Mester wrote that the findings indicate persistent informational and structural asymmetries between large and small digital-asset investors — an interpretation, not an independent regulatory conclusion.

Same-Direction Trades Do Not Establish Intent

A cohort trading in a whale's direction after an alert is an association, not an observed decision to copy. The event study cannot confirm that any wallet operator saw the alert, and CryptoSlate's coverage explicitly cautioned that the analysis shows associations rather than proof that notifications caused trades. Whale status in the study is also a transaction-based proxy, defined as any wallet that transferred more than $50 million at least once, so it is not a verified identity or a current-balance classification.

Why the Finding Alone Is Not a Trading Signal

Wallets are not people. One wallet does not equal one trader, and nothing in the participation-share data identifies these cohorts as retail investors, market makers or automated strategies. The paper establishes no profitability, no causation and no forward price direction, and framings suggesting that whale alerts let holders profitably copy trades go beyond what the evidence supports.

The distinction matters for anyone parsing on-chain whale narratives. When a dormant whale moves billions, the reflex is to read intent into the flow; this paper suggests the observable footprint is a short-lived cohort reaction, not a verified copy-trade. As broad market context rather than any reaction to the paper, Bitcoin traded at $77,303 with the Fear & Greed Index at 63, or "Greed."

Study Details Needed to Put the Finding in Context

The event-isolation design keeps the samples tight. Restricting events to those without another whale transaction within 120 minutes on either side leaves 6,645 Bitcoin and 5,075 Ethereum whale transactions, drawn from Whale Alert notifications spanning December 14, 2017 through December 31, 2025.

Definitions, Sample, and Effect Size

The directional regressions run on a narrower set than the isolated event counts. After dropping observations with unclear direction, the buy/sell analysis uses 5,884 Bitcoin events — 3,471 buys and 2,413 sells — and 4,843 Ethereum events. That directional sample count bounds what the cohort-level claims can support.

Non-whale size groups are reassigned during fixed January-June and July-December intervals using each wallet's largest transaction: small below the 50th percentile, medium from the 50th to 95th percentile and large at or above the 95th percentile, while whale status stays fixed throughout. That the medium cohort shows the largest swings is partly a function of this percentile banding, not evidence about individual behavior. The same caution applies to any single-wallet event, from this study's cohorts to the ETF flow swings that headline daily coverage.

Readers should also weigh the document's status. The Philadelphia Fed is one of the 12 regional Reserve banks of the Federal Reserve System, and its working papers are discussion drafts whose figures can change in later versions. WP 26-42 identifies itself as preliminary research circulated for discussion, and its disclaimer states that the authors' views do not necessarily reflect those of the Philadelphia Fed or the Federal Reserve System. It is central-bank research on crypto market structure — not a rule, enforcement action or adopted policy — and no new investor obligation follows from it.