BtcLiquidationCascadeExhaustionReversalLS15m

BTC Liquidation-Cascade Exhaustion Reversal, Long-Short (Binance USD-M, 15m Bars, Forced-Flow Overshoot Fade, ATR-Stop, Mean-Revert-to-VWAP Target, Low-Parameter)

A LONG-SHORT, single-instrument, event-driven MEAN-REVERSION strategy on BTCUSDT.BINANCE USD-M perpetual using 15-MINUTE bars, triggered by LIQUIDATION-CASCADE EXHAUSTION rather than price momentum. The mechanism is mechanistically ORTHOGONAL to the factory's continuation/breakout/momentum families (which are dead-or-marginal on BTC): instead of betting that a move persists, it fades the transient price OVERSHOOT created by forced, price-insensitive liquidation flow. When a wave of long liquidations (forced selling) spikes far above its trailing norm AND price is simultaneously stretched to an oversold extreme, the forced sellers are exhausting non-informational supply below fair value — the strategy buys the snap-back. Symmetrically, a short-liquidation (forced-buying) spike that stretches price to an overbought extreme is faded short. Holds are intraday (minutes to a few hours), targeting the 1-3% reversion to the local mean — an order of magnitude larger than the ~0.10% round-trip fee, so this is NOT a fee-marginal microstructure scalp and NOT the cross-venue basis trap. Deliberately Binance USD-M because that is where multi-year, reliable liquidation data exists (sidestepping Hyperliquid's short-history wall); the long-short + short-horizon design is chosen to attack the portfolio's two worst quota gaps (85.5% long-only, 3.7% short-horizon). Kept to 4 parameters to resist the overfitting that has recurred across the recent batch.

data_unavailable: backtest data for ? could not be collected after 5 retries: - supplementary:liquidations: got 0, needed 1 The data layer (not the strategy) is the blocker — re-evaluate this hypothesis once collection is fixed / the symbol is backfilled.

CLEAN RESTART 2026-09-04 — this run's verdict history and learning records were removed and it was restarted from verification. Its previous abandonment came from the pipeline, not from the market: the Layer-2 harness mis-bound @staticmethod helpers (fixed), QA issued terminal performance verdicts on an unoptimized smoke test (removed — QA now judges correctness only), and sandbox timeouts came from backtest-slot starvation (fixed). The hypothesis and the strategy code are unchanged. Verify the code on its merits; performance is decided later by the full backtest and the optimizer.