Data · Hyperliquid × Bybit · snapshot 2026-08-06 18:15 UTC · refreshed every ~2 h
How much whale money is long — and how many retail accounts are long in the same minute. Whale positions are read wallet by wallet from the Hyperliquid chain; retail comes from Bybit. When the two series diverge, one side is wrong, and it is rarely the side with the billions.
For whales the base is notional: how many dollars sit long versus short. For retail the base is accounts: how many people sit long versus short. One whale with a billion and a thousand retail traders with a hundred each produce opposite pictures from the same data.
That is why we don't compute a single «crowd vs whales index». It would look more convincing and it would be false. We show two series and state the gap in points without pretending the two are commensurable.
The second thing worth knowing: whales on Hyperliquid are structurally short — the median long share for BTC across our archive is 26%. Part of that short book hedges spot rather than betting on a fall. So «whales are short» is not a bearish signal by itself; what matters is the deviation from their own norm. Across our entire archive whales have never once been majority long: the highest BTC long share on record is 45%. So a high percentile reads as «least short on record», not «flipped long».
📍 Зараз: BTC whales are 38% long against 54% of retail accounts; for ETH it is 31% against 65%.
Whales — ~30-minute snapshots from our own archive; retail — Bybit's 4-hour series. Spot is off by default, toggle it on.
| Series | What is measured | Source | Step · depth |
|---|---|---|---|
| Whales | long share of top-whale notional, BTC/ETH | Hyperliquid, on-chain positions per wallet | ~30 min · since Jul 10, 2026 |
| Retail | share of accounts long | Bybit v5 account-ratio (global) | 4 h · since May 15, 2026 |
Whale long share = long notional ÷ (long notional + short notional) at each snapshot. Snapshots with fewer than 40 polled holders are dropped — a partial poll skews the picture.
The percentile is where the current value sits among all our snapshots: 90 means it has been higher only 10% of the time across the whole archive. Same mechanic as in our options fear index.
Bybit serves only the last 500 points as a rolling window — at a 4-hour step that is 83 days, and every day the earliest one drops out of the API for good. So we merge each fresh window into our own archive: points stay with us after the exchange stops serving them. The left part of the chart carries a single series — retail already exists there, our whale archive does not.
Want the divergence spotted for you — Sentinel carries a «crowd vs whales» alert: plans and setup.
Data: Hyperliquid (on-chain) · Bybit v5 API. Refreshed every ~2 hours.
Educational material and a journal of observations. Not investment advice, not a solicitation to buy or sell. Past market behaviour guarantees nothing.