Q3 2026 Ethereum Whale Flow Report: Full-Quarter Recap (Jul–Sep)
Whale Sentiment Index by month, whale DEX swaps, and exact exchange flow for September 6–30, with a plain account of what the data can and can't measure.
Published 2026-09-09 · Updated 2026-10-05 · Deep Blue Alpha
Across Q3 2026 (July 1 to September 30), the Deep Blue Alpha Whale Sentiment Index averaged 49.3 out of 100, close to neutral. It averaged 52.6 in July, 46.4 in August and 49.0 in September, read 50 or higher on 40 days and below 50 on 52 days, peaked at 89 on August 26 and closed the quarter at 21 on September 30.
Exact whale-scoped exchange flow exists only for September 6–30, 2026, the last 25 days of the quarter. In that window, tracked whale wallets withdrew $688.8M of directional tokens from exchanges and deposited $340.2M, a net +$348.6M leaving exchanges and a 66.9% withdrawal share. UNI led by gross flow at $216.6M; LINK had the largest net figure at +$45.12M.
Most whale activity was exchange movement, not trading: from September 6 to 30, exchange transfers were 97.4% of all whale-scoped dollar value and DEX swaps 2.6%. Whale DEX swaps rose from 7,427 in July to 12,960 in September, though the tracked wallet set changed during the quarter, so part of that change reflects coverage.
What changed in this version of the report?
The first version of this page, published on September 9, was a single-day snapshot with a quarterly label. It also mixed whale wallets with the exchange side of every transfer, which inflated the totals. Now that Q3 2026 has closed, the page has been rebuilt as a full-quarter recap, and every number in it is whale-scoped: it counts activity by wallets in the Deep Blue Alpha tracked set (21,889 wallets on October 5, 2026), excludes self-transfers, and drops any single row valued at $1B or more as a pricing-error guard.
The result is a smaller, more honest set of numbers. Instead of one dramatic daily total, the report covers three things: how the Whale Sentiment Index moved across all three months, how much tracked whales swapped on DEXes each month, and a detailed look at September 6–30, the one stretch of the quarter where exact exchange flow exists at whale scope.
What can this report measure, and what can't it?
- Exact exchange flow covers September 6–30 only. Raw exchange-transfer rows are kept for 30 days on a rolling basis. When the data was pulled on October 5, rows from September 6 onward were still held, so withdrawals and deposits for tracked whale wallets can be measured precisely for the last 25 days of September only.
- July, August and September 1–5 exchange flow is not covered. This report does not estimate it, back-fill it, or substitute another source. Those periods are covered only through the Whale Sentiment Index and DEX swaps.
- The Whale Sentiment Index and DEX swaps cover all of Q3. Both are stored for the full quarter, so they are the only like-for-like series from July through September.
- The tracked wallet set changed during the quarter. Admission moved to a holdings rule (at least $250K in real holdings, stablecoins excluded), and market-maker, exchange and desk wallets were removed. Month-over-month comparisons therefore partly reflect who was being tracked, not only what whales did.
- Totals use the wallet set as of October 5. Wallets admitted or removed during September shift the September 6–30 totals, so week-to-week changes inside that window are not presented as a trend.
These limits are stated up front because they change how the numbers can be read. A quarterly whale report that quietly blends different wallet sets, or fills a missing month with a guess, looks more complete than it is. This one shows what was measured and labels what was not.
How did the Whale Sentiment Index move across Q3 2026?
The Whale Sentiment Index is a daily 0–100 score built from tracked whale wallet activity. Readings above 50 mean the day leaned toward the accumulation side (more value leaving exchanges than arriving); readings below 50 mean it leaned toward the distribution side. It is a summary of observed past flow, not a forecast. The full method is explained in What Is the Whale Sentiment Index?
| Month | Average | Range | Days ≥ 50 | Days < 50 | High | Low |
|---|---|---|---|---|---|---|
| July | 52.6 | 17–85 | 17 | 14 | 85 (Jul 16) | 17 (Jul 22) |
| August | 46.4 | 20–89 | 11 | 20 | 89 (Aug 26) | 20 (Aug 1) |
| September | 49.0 | 17–76 | 12 | 18 | 76 (Sep 23) | 17 (Sep 13) |
| Q3 total | 49.3 | 17–89 | 40 | 52 | 89 (Aug 26) | 17 (Jul 22, Sep 13) |
Daily Whale Sentiment Index scores, July 1 to September 30, 2026 (92 days). Monthly day counts computed from the daily series.
July: the only month above 50
July opened in a narrow band, with the index reading between 55 and 58 for the first six days. On July 7 it jumped to 83 and stayed between 70 and 85 for the next ten days, reaching the quarter's July high of 85 on July 16. The next day it fell to 47, and by July 22 it had dropped to 17, the lowest reading of the quarter. The rest of the month stayed low, between 18 and 31. That split, a strong first half and a weak second half, is why July still averaged 52.6 and finished with 17 days at or above 50.
August: mostly below 50, with a late spike
August started at 20 on August 1 and spent most of the month in the 30s and 40s. Only 11 of its 31 days read 50 or higher, the fewest of the quarter. The exception was the final week: the index climbed from 48 on August 24 to 65 on August 25 and then to 89 on August 26, the highest reading of Q3. It eased to 62 and 58 over the next two days and closed the month at 77 on August 31. Even with that run, August averaged 46.4, the lowest monthly average of the quarter.
September: wide swings around neutral
September averaged 49.0, almost exactly neutral, but the daily path was uneven. The index matched the quarter low of 17 on September 13, then recovered to 59 the next day. A run between September 20 and September 23 produced readings of 70, 66, 61 and 76, with 76 the month's high. The last two days of the month read 34 and 21, so the quarter closed at 21 on September 30. In total, 12 September days read 50 or higher and 18 read below it.
The first four days of October read 44, 40, 11 and 56. Those are outside Q3 and are listed only to show where the series stood when this recap was written. Daily readings move quickly, so the live Whale Sentiment Index page is the place to see the current score.
How many DEX swaps did tracked whales make each month?
DEX swaps are the second series that covers the whole quarter. These are actual trades on decentralized exchanges by tracked whale wallets, across all non-stablecoin tokens including ETH and WETH. Because a swap is an observed exchange of one token for another, it is the one place in this report where buy-side and sell-side language describes what actually happened.
| Month | DEX swaps | Wallets | Buy-side $ | Sell-side $ |
|---|---|---|---|---|
| July | 7,427 | 198 | $26.2M | $20.4M |
| August | 10,957 | 310 | $188.4M | $133.2M |
| September | 12,960 | 255 | $148.2M | $139.7M |
Whale-scoped DEX swaps, all non-stablecoin tokens including ETH/WETH.
Swap counts rose in each month of the quarter, from 7,427 in July to 10,957 in August and 12,960 in September. The dollar value moved differently. July was small on both sides, with $26.2M on the buy side and $20.4M on the sell side. August was the largest month by value, with $188.4M buy-side against $133.2M sell-side, and it also had the most distinct wallets swapping, at 310. September had the most swaps but fewer wallets (255), and its two sides were closer together, at $148.2M and $139.7M.
In all three months, buy-side swap value was higher than sell-side swap value. That is an observation about the quarter's swaps, not a statement about direction from here. It also needs the coverage caveat above: because the tracked wallet set changed during Q3, part of the jump from July to August reflects which wallets were being followed. The whales buying today board separates live DEX swaps from exchange flow for anyone who wants to see the current split.
Why do September 6–30 get their own deep dive?
September 6–30 is the one stretch of Q3 with exact whale-scoped exchange flow, because raw exchange-transfer rows are kept for 30 days. Those 25 days also show how much the scoping choice matters. Deep Blue Alpha recorded 1,684,170 exchange-transfer rows from September 6 to 30, worth $182.0B. Only 20,354 of those rows (1.21%), worth $9.49B, belonged to tracked whale wallets. The other 98.8% were the exchange side of transfers and wallets outside the tracked set.
That gap is the reason the old version of this page was replaced. A raw total counts both ends of a transfer and every wallet that touched an exchange. A whale-scoped total counts only what tracked whale wallets did. Everything in the September 6–30 sections below uses the whale-scoped figures.
What did whale activity look like from September 6 to 30 by asset group?
Whale-scoped activity from September 6 to 30 totaled $9,742.6M across 31,523 moves. It splits into three asset groups, and each one behaved differently.
| Asset group | Moves | Value | Exchange share of $ | DEX share of $ | DEX share of moves |
|---|---|---|---|---|---|
| ETH / WETH / wBTC | 14,832 | $4,004.7M | 93.8% | 6.2% | 61.6% |
| Directional tokens | 11,637 | $1,037.9M | 99.1% | 0.9% | 17.5% |
| Stablecoins | 5,054 | $4,700.0M | 100.0% | 0% | 0% |
| All | 31,523 | $9,742.6M | 97.4% | 2.6% | 35.4% |
September 6–30, 2026. Directional tokens exclude stablecoins, ETH/WETH and wrapped BTC. Gold-pegged XAUT and PAXG are excluded from token rankings.
The pattern is consistent across groups: DEX swaps are frequent but small, and exchange transfers carry the value. For ETH, WETH and wrapped BTC, swaps were 61.6% of moves but only 6.2% of value. For directional tokens, swaps were 17.5% of moves and 0.9% of value. Stablecoin activity was entirely exchange movement. This is why the vocabulary in this report is careful: calling this window's whale flow "buying" or "selling" would describe trades that, for the most part, never happened. The companion study, Exchange Withdrawals vs DEX Buys, explains the split in more detail.
How much directional-token value left exchanges from September 6 to 30?
Directional tokens are the closest thing in this dataset to a view on specific projects, because they exclude stablecoins and the ETH/BTC base layer. From September 6 to 30, 789 tracked whale wallets made 11,637 directional-token moves. Of that activity:
- $688.8M was withdrawn from exchanges (tokens moving from an exchange to a whale wallet, the accumulation-side read).
- $340.2M was deposited to exchanges (tokens moving from a whale wallet to an exchange, the distribution-side read).
- That leaves a net +$348.6M leaving exchanges, with a 66.9% withdrawal share.
- DEX swaps were small by comparison: $5.7M buy-side and $3.2M sell-side.
In other words, directional-token whale flow over those 25 days leaned to the accumulation side by roughly two dollars out for every one dollar in. The heuristic behind that label is simple: coins leaving exchanges are read as accumulation-side, coins arriving are read as distribution-side. It does not reveal why any individual wallet moved its tokens. A withdrawal can precede staking, lending, bridging or cold storage just as easily as long-term holding.
Where did the value sit by move size?
| Move size | Moves | Value | DEX share of moves | DEX share of $ |
|---|---|---|---|---|
| Under $10K | 6,502 | $19.5M | 29.3% | 21.4% |
| $10K–$100K | 3,325 | $128.1M | 3.8% | 2.2% |
| $100K–$1M | 1,638 | $494.5M | 0.2% | 0.4% |
| $1M–$10M | 172 | $395.7M | 0.0% | 0.0% |
September 6–30, 2026, directional tokens, whale-scoped. No single directional-token move reached $10M.
More than half of all directional-token moves (6,502, or 55.9%) were under $10K, but together they added up to just $19.5M. The value sat in the middle and upper bands: 1,638 moves between $100K and $1M carried $494.5M, and only 172 moves between $1M and $10M carried $395.7M. DEX swaps showed up almost entirely in the smallest band, where they were 29.3% of moves. Above $100K, they were close to zero. Large whale moves on directional tokens were exchange transfers.
Which tokens led whale exchange flow from September 6 to 30?
The table ranks directional tokens by gross whale-scoped flow from September 6 to 30. Net leaving exchanges is withdrawals minus deposits; withdrawal share is withdrawals divided by withdrawals plus deposits.
| Token | Moves | Wallets | Gross | Withdrawn | Deposited | Net leaving exch. | Withdrawal share | DEX % |
|---|---|---|---|---|---|---|---|---|
| UNI | 990 | 132 | $216.6M | $129.18M | $87.35M | +$41.83M | 59.7% | 0.0% |
| LINK | 593 | 126 | $116.4M | $80.67M | $35.55M | +$45.12M | 69.4% | 0.1% |
| QNT | 569 | 69 | $93.1M | $65.59M | $27.40M | +$38.20M | 70.5% | 0.1% |
| ONDO | 490 | 72 | $81.6M | $59.62M | $21.93M | +$37.69M | 73.1% | 0.1% |
| ENA | 623 | 67 | $68.8M | $46.15M | $22.58M | +$23.57M | 67.1% | 0.2% |
| PEPE | 274 | 62 | $36.5M | $22.38M | $14.07M | +$8.31M | 61.4% | 0.2% |
| AAVE | 326 | 80 | $30.4M | $19.42M | $10.92M | +$8.50M | 64.0% | 0.3% |
| WLD | 138 | 32 | $28.4M | $23.06M | $5.36M | +$17.69M | 81.1% | 0.0% |
| SKY | 109 | 30 | $21.4M | $10.05M | $11.30M | −$1.25M | 47.1% | 0.1% |
| FET | 190 | 36 | $16.8M | $11.84M | $4.95M | +$6.88M | 70.5% | 0.0% |
| SHIB | 196 | 44 | $16.6M | $12.62M | $3.97M | +$8.64M | 76.0% | 0.0% |
| WLFI | 115 | 21 | $15.7M | $8.06M | $7.66M | +$0.40M | 51.3% | 0.0% |
September 6–30, 2026, directional tokens, whale-scoped. XAUT and PAXG (gold-pegged) removed. BNT ($18.5M gross, 82 moves) is omitted because every move came from a single wallet.
What stands out in the token table
UNI led by size, LINK led by net. UNI had the most gross flow at $216.6M and the most moves (990) and wallets (132). Because its deposits were also large ($87.35M), its withdrawal share was 59.7%, among the lower readings in the table. LINK had about half of UNI's gross flow but the largest net figure, +$45.12M leaving exchanges, from 126 wallets.
QNT and ONDO had high withdrawal shares on fewer wallets. QNT showed +$38.20M net leaving exchanges at a 70.5% withdrawal share from 69 wallets, and ONDO +$37.69M at 73.1% from 72 wallets. Both moved about as much net value off exchanges as UNI did, with roughly half as many wallets involved.
WLD had the highest withdrawal share. At 81.1%, WLD's $23.06M in withdrawals was more than four times its $5.36M in deposits, though from a smaller group of 32 wallets.
SKY was the only token in the table with net deposits. Whale wallets deposited $11.30M to exchanges and withdrew $10.05M, a net −$1.25M and a 47.1% withdrawal share, a mild distribution-side lean. WLFI was close to balanced at +$0.40M.
DEX swaps barely registered. No token in the table had more than 0.3% of its whale-scoped value in DEX swaps. These are exchange-flow readings, and the labels used here reflect that.
For current figures on any of these tokens, the live pages for UNI, LINK, QNT, ONDO and ENA update continuously.
What did whales do with stablecoins on exchanges from September 6 to 30?
Stablecoins were the largest single block of whale value from September 6 to 30. They accounted for 5,054 moves and $4,700.0M, 48.2% of all whale-scoped dollar value. Stablecoin moves were also much larger than everything else: the median stablecoin move was $100K, against $5.9K for non-stablecoin moves (about 17×), and the average was $930K against $191K.
| Stablecoin | Withdrawn | Deposited | Moves | Wallets |
|---|---|---|---|---|
| USDC | $2,200.8M | $415.6M | 2,047 | 334 |
| USDT | $1,345.7M | $687.2M | 2,935 | 430 |
| PYUSD | $44.3M | $0 | 26 | 5 |
| DAI | $0.9M | $4.6M | 5 | 4 |
| USDe | $0.8M | $0.1M | 35 | 8 |
September 6–30, 2026, whale-scoped stablecoin exchange transfers.
USDC dominated. Tracked whale wallets withdrew $2,200.8M of USDC from exchanges against $415.6M deposited, a net +$1,785.2M leaving exchanges and an 84.1% withdrawal share, across 2,047 moves by 334 wallets. The average USDC move was $1.28M, and the largest single move was $50.0M. USDT showed the same direction at a smaller ratio: $1,345.7M withdrawn against $687.2M deposited (net +$658.5M, a 66.2% withdrawal share), across 2,935 moves by 430 wallets, with an average move of $693K and a largest single move of $44.0M. PYUSD was one-sided, with $44.3M withdrawn and nothing deposited, but from only five wallets.
Stablecoin flow is harder to label than token flow. A stablecoin leaving an exchange can be headed to DeFi lending, to an OTC settlement, to another exchange, or into storage. It is not, by itself, a directional read on any asset. That is why stablecoins sit in their own section and are excluded from the directional-token totals above. The same direction continued into the first days of October: from October 1 to 4, USDC withdrawals were $274.2M against $74.0M deposited (319 moves, 111 wallets), and USDT withdrawals were $255.5M against $91.7M deposited (451 moves, 148 wallets).
More context on how whales move stablecoins on-chain is in How Whales Use Stablecoins.
How is whale exchange flow read correctly, and what can't it show?
Every September 6–30 figure in this report is labeled with exchange-flow language, on purpose. The rules are short:
- A withdrawal is not a purchase. It means tokens moved from an exchange to a tracked whale wallet. The trade, if there was one, happened inside the exchange and is not visible on-chain.
- A deposit is not a sale. It means tokens moved from a whale wallet to an exchange. What happened next is not observable.
- Accumulation-side and distribution-side are descriptive labels. They summarize which way past flow leaned, using the heuristic that coins leaving exchanges read as accumulation and coins arriving read as distribution.
- Only DEX swaps are observed trades. From September 6 to 30, they were 2.6% of whale-scoped value and 0.9% of directional-token value.
- Check breadth. A figure from one wallet describes that wallet. BNT's $18.5M from September 6 to 30 came from a single address, which is why it is omitted from the token table.
The companion data study, Exchange Withdrawals vs DEX Buys: What Whale Flow Really Is, goes deeper into why roughly 97% of whale flow is exchange transfers and what that means for anyone citing whale data. For background on exchange flows in general, see Crypto Exchange Inflows and Outflows Explained.
Bottom line
Q3 2026 was close to neutral on the Whale Sentiment Index. The quarter averaged 49.3, with July above 50 at 52.6, August below it at 46.4, and September almost exactly in the middle at 49.0. The extremes were brief: 89 on August 26 and 17 on July 22 and September 13. The quarter closed at 21 on September 30.
September 6–30, the only stretch of the quarter with exact whale-scoped exchange flow, leaned to the accumulation side for directional tokens: $688.8M withdrawn from exchanges, $340.2M deposited, a net +$348.6M leaving exchanges and a 66.9% withdrawal share. UNI led by gross flow, LINK by net, and SKY was the only top token with net deposits. Stablecoins carried nearly half of all whale-scoped value, with USDC withdrawals of $2,200.8M against $415.6M deposited.
The most important finding is about the data itself. From September 6 to 30, 97.4% of whale-scoped value was exchange transfers and 2.6% was DEX swaps. Whale flow is mostly coins moving to and from exchanges, and it is described here in those terms. The report also states where its coverage ends: exact exchange flow only for September 6–30, and a tracked wallet set that changed during the quarter.
Data source: Deep Blue Alpha whale-scoped on-chain data for Q3 2026 (July 1 to September 30), pulled October 5, 2026; exchange-flow figures cover September 6–30, 2026 only. Whale-scoped means activity by wallets in the tracked set as of October 5, 2026 (21,889 wallets), excluding self-transfers and rows of $1B or more. Past whale activity is not predictive of future activity. NFA / DYOR.
See whale exchange flow as it happens
Deep Blue Alpha tracks 20,000+ Ethereum whale wallets, with live exchange flow, DEX swaps, token-level breakdowns and the daily Whale Sentiment Index — the same dataset behind this report.
Open today's whale board →