A LONG-SHORT, single-instrument reversal on BNBUSDT.BINANCE (USD-M perp, 1h bars) built on an ORDER-FLOW ABSORPTION signal from the taker buy/sell volume feed (takerBuySellVolume, catalogued multi-year via Binance metrics) — a data source none of my prior proposals or the in-pipeline set uses. The mechanism is the classic Wyckoff 'effort vs result' divergence: when aggressive market SELLING spikes to an extreme (taker-sell volume dominates) yet price REFUSES to make a lower low and closes in the upper part of its bar, the forced selling is being absorbed by hidden resting demand — a bullish tell — so go LONG; mirror for a taker-BUY spike that fails to make a higher high (supply absorption) → SHORT. This is deliberately distinct from every dead/in-pipeline family: NOT price/VWAP z-reversion (the trigger is a flow/price DIVERGENCE, not a price deviation), NOT the volume-spike capitulation flush (that fired on a raw volume z-spike with a fixed reversion take-profit that clipped winners; this fires on directional TAKER-flow absorption and uses a symmetric ATR exit), NOT crowding fades on OI/premium/L-S ratios (different feed — aggressor flow), NOT liquidation-feed dependent (multi-year taker feed, no 90-day wall), NOT breakout/trend, and NOT funding-carry or cross-sectional rotation. It adds to the under-represented long-short bucket (14% vs quota) with a fresh microstructure feed.