How I Built a Real-Time Bitcoin Liquidation Heatmap From Scratch

Every liquidation heatmap I found was lying to me. Not intentionally. They were just slow. Two-minute update intervals in a market where $500M gets liquidated in 30 seconds. By the time the map refreshed, the move was already over. The data was a postmortem, not a tool. So I built one that updates every 12 seconds, pulling from four exchanges simultaneously. Here is exactly how. Architecture: Four Exchanges, One Map Binance ─┐ Bybit ─┤──▶ Aggregation Layer ──▶ Normalization ──▶ Canvas Render OKX ─┤ (merge) (weight) (draw) Aster ─┘ Each exchange exposes liquidation and order book data through WebSocket feeds and REST endpoints. I poll all four in parallel on a 12-second interval, merge the results into a single dataset, normalize across exchanges, and paint it onto a canvas element. Why 12 seconds? It is the shortest interval I could sustain across all four APIs without getting rate-limited. Binance is the bottleneck. The Hard Part: Cross-Exchange Normalization Binance reports 10x the v