Ethereum Whale Flow Dataset: Labeled Whale Moves as CSV (Free Sample)

Most “whale datasets” online are one-off exports: a list of large transfers or top holders at one moment, with no labels and no updates. A dataset meant for research or modeling needs a stable wallet definition, clear labels, a documented history start and a way to keep it current. This page describes Deep Blue Alpha’s whale-flow dataset on those points and links the free sample so you can check it directly.

Columns

The first rows of the Oct 16, 2026 sample file:

time_utc,tx_hash,wallet,token,action,usd_value,venue_type
2026-10-16T10:41:23Z,0x8b85a18c…42b93,0x83d86484…0e09e,WETH,SWAP_SELL,4539.42,dex
2026-10-16T10:41:23Z,0xfe711607…5a2,0xe0554a47…939f,WETH,SWAP_BUY,2286.03,dex
2026-10-16T10:41:11Z,0xe05ab113…e680,0x0cd6db9c…66fc,WETH,CEX_DEPOSIT,125376.34,exchange

(Hashes shortened here; the file has full values.)

How to load the sample

1. Download the sample. Get https://deepbluealpha.io/data/sample.csv (about 0.8 MB, last 7 days, top 10 coins by tracked-whale volume).
2. Load it, skipping the comment line. Line 1 starts with # and the header is line 2: pd.read_csv(url, comment='#').
3. Split by venue. Group by token and venue_type to see exchange flow and DEX swaps separately.
4. Join on-chain detail if needed. Use tx_hash to look up any row on a block explorer.
5. Get full history when Business opens. Daily CSV + JSON files (one per UTC day, from Oct 7, 2026) are launching soon on the Business plan; the API is available on Whale today for personal use.

import pandas as pd
df = pd.read_csv("https://deepbluealpha.io/data/sample.csv", comment="#")
df.groupby(["token", "venue_type"])["usd_value"].sum()

What you can measure with it

• Exchange flow per token and day: filter venue_type == "exchange", then sum CEX_WITHDRAW minus CEX_DEPOSIT by token and date. Coins leaving exchanges are commonly read as the accumulation side; coins arriving as the distribution side.
• DEX trading by whales: filter venue_type == "dex"; these are the only rows that are trades.
• Flow mix: the share of rows or dollars that are exchange transfers vs swaps. In the Oct 16 sample, 3,517 of 4,536 rows (78%) were exchange moves.
• Wallet concentration: group by wallet to see how much of a token’s flow came from a few wallets. A few very active wallets can account for many rows.
• Event studies: join on time_utc with your own price or event data. Note the short history (from Oct 7, 2026) before drawing conclusions.

Two cautions for analysis. The sample’s top 10 coins can include stablecoins when they rank by volume, so exclude them if you only want volatile tokens. And totals from different weeks describe slightly different wallet groups, because the tracked set is re-checked every few hours as wallets qualify or drop out.

Sample vs full data

Other places to get Ethereum whale data in bulk

Whale Alert’s Enterprise Big plan (from $2,000/mo) offers full history on 12 chains with bulk JSON/Parquet exports of large transfers; Enterprise ($699/mo) includes 90 days. Dune lets paid plans (Analyst $75/mo, Plus $399/mo) export query results, billed in credits per MB. CryptoQuant and Glassnode provide metric time series rather than row-level whale moves. Sources: developer.whale-alert.io; Dune API billing, Dune rate limits; checked Oct 16, 2026.

Known limits

• History is short: Oct 7, 2026 onward for tracked whales. Earlier DEX history exists for some wallets but is not uniform, so it is not offered as a dataset.
• Labels are heuristics from known address lists; a new or unlabeled venue can be missed or mislabeled. Wallet labels never identify a person.
• Coverage is the tracked set, not every Ethereum wallet, and the set changes over time.
• Ethereum mainnet only.