Bybit ETH Liquidation-Absorption Breakout with Fixed 24-Hour Exit

Hypothesis

A single-leg, long-short ETHUSDT.BYBIT strategy on 4-hour bars. Remain flat unless a large directional displacement occurs on unusually high volume and closes near the bar extreme, indicating forced-flow absorption. Enter next bar open and exit after exactly six 4-hour bars; expected flat time is approximately 90%.

Preserves the prior displacement, volume, and close-location signal while adding bounded point-in-time liquidation confirmation through supp_window("liquidations", 64), filtered by timestamp_ns, usd_value, bar.ts_init, and a four-hour age limit.

data_unavailable: Layer 3 (sandbox backtest) could not run 3 times — No bar data found for ETHUSDT.BYBIT-4-HOUR-LAST-EXTERNAL in catalog at data/catalog This is a pipeline/data-layer failure, not a verdict on the strategy. Revive the run once the cause is fixed.

Implementation

Trades ETHUSDT on Bybit after a high-volume directional displacement closes near an extreme and is confirmed by recent same-direction forced liquidations, then exits after six 4-hour bars.

Iteration History

Verification failed (Layer 4 — QA review) [class=hypothesis_mismatch]: - [edge_concern] The smoke test reports a positive 0.2115% average trade return, but impact cost is reported as 39.9058% and commission as 7.22% of gross; economics should be rechecked after correcting the signal. - [critical] The hypothesis's core edge is liquidation absorption, but the strategy never reads liquidation data or any liquidation-derived supplementary series. It only combines OHLCV displacement, volume, and close location, implementing a volume/displacement breakout instead. (line 53) — fix: Read liquidation data through supp_as_of() or supp_window(), bounded at bar.ts_init, and require the appropriate directional forced-flow condition in the entry signal.

Outcome Summary

Ensure the required Bybit 4-hour bar data is available before retrying the backtest.

The run was abandoned after Layer 3 failed three times because the required Bybit bar data was missing from the catalog. This was recorded as a data-layer failure, not a strategy verdict; later stages were not reached.

The strategy tried to trade ETHUSDT on Bybit after large, high-volume directional bars with near-extreme closes and matching liquidation flow, exiting after six 4-hour bars.

No backtest metrics, returns, Sharpe, or trade counts are available because the sandbox backtest could not run.