A long-short mean-reversion strategy on BTCUSDT.BINANCE USD-M perpetual futures that fades EXTREME aggressive-flow imbalances. Uses BINANCE USD-M perp bars (proven clean) + taker_ratio supplementary data (PostgreSQL — confirmed available, completely unused across 464 prior experiments). The hypothesis: when aggressive taker BUY volume dominates the 4h rolling window by >2.5 standard deviations vs the 30-day baseline, retail FOMO buying has exhausted — the marginal buyer who was willing to pay-up has been satisfied, and price reverts as natural sellers and short-side liquidity-providers step in. Symmetric on the SHORT side: when aggressive taker SELL volume dominates by ≤-2.5 SD, panic-selling has exhausted and a reversion follows. This mechanism is STRUCTURALLY DISTINCT from the recently-documented 'BTC funding-extreme mean-reversion decay post-2023' failure pattern: funding-rate extremes are POSITIONING-state lagging indicators that institutional capital can preempt; taker-flow extremes are REAL-TIME AGGRESSION-flow measurements that can only be observed AFTER the flow has happened, making them harder to arbitrage and more durable across regimes. Fills FIVE under-represented buckets: long-short direction (14.0% → toward 45%), taker_ratio-data-driven mechanism (zero of 464 experiments use this stream), aggressive-flow-microstructure edge (distinct from positioning, funding, liquidations, L/S), mid-term hold (8-24h cycles), and BINANCE-USD-M-only (clean data, avoiding all BINANCE_SPOT failure modes). Position sizing: 15% of equity per direction. Risk per trade: 1.5% of equity capped by stop-loss.