Nvidia Liquidation Heatmap: $337.6K in 24h

Over the last 30 days, $7.3M in NVDA positions was liquidated on Hyperliquid across 23 days with liquidations — about $316.9K per day. The heaviest day was 2026-08-26, with $4.5M force-closed — roughly 14× a typical day. The latest complete day landed above that average.

Longs absorbed 81% of the 30-day damage: most NVDA liquidations came on moves down.

On 2026-08-27, NVDA ranked #15 among Hyperliquid markets by liquidated notional — 1% of the day's total. All markets side by side live on the liquidations hub; for contrast, see SPCX liquidations.

The map at the top plots price and time against liquidation density. Each horizontal band marks a price level where leveraged positions would be force-closed; the brighter the band, the more notional is at risk there. Bands persist until price trades through them — a band that ends against a candle means those positions were flushed.

Clusters act like magnets. A dense band above or below the current price is forced buying or selling waiting to happen: when price reaches it, liquidations execute at market and push price further in the same direction. The right-hand profile sums this up — how much would be liquidated on a move up (shorts) or down (longs) from the current price.

The threshold slider clips the colour scale at a percentile: at 0.85 everything above the 85th percentile of density burns at full brightness, which makes the big clusters stand out. Candles and the daily totals below the map are settled on-chain data; the heat bands are currently a model estimate built from price and volume at typical leverage tiers (100×, 50×, 25×, 10×, 5×) — the same approach other heatmaps use. Unlike them, Hyperliquid settles everything on-chain, so the next step here is replacing the estimate with actual liquidation prices of real wallets.

A liquidation happens when a leveraged position can no longer cover its margin and the exchange closes it at market. The number above is the notional value of positions force-closed in the last complete day — not trader losses, which are smaller and depend on entry price.

Longs and shorts tell opposite stories. Longs get liquidated when price falls; shorts when it rises. A day dominated by short liquidations usually means a sharp rally caught bearish positioning offside — and the forced buying pushes price further up, which is why these events cluster.

Why Hyperliquid numbers differ from centralised exchanges. On a CEX, liquidation data is whatever the venue chooses to publish, and most heatmaps you find elsewhere are estimates reconstructed from order books. Hyperliquid settles every fill on-chain, so these totals are counted from settled trades rather than modelled. The trade-off: this covers Hyperliquid only, not the whole market.

New to perpetuals? Start with what Hyperliquid is and how on-chain perps work.

Looking for who is trading these moves? See the Hyperliquid leaderboard for the traders behind the volume, or markets for per-coin activity.