Hyperliquid LTC–BCH Relative Shock Exhaustion

Hypothesis

Test whether unusually large relative moves between two established proof-of-work assets partially reverse after observable exhaustion. Trade equal-dollar, opposing LTC and BCH perpetual positions using synchronized 4-HOUR bars. HyperliquidCollector supplies both exact perpetual instruments. Local stamped catalog coverage overlaps from 2024-01-25T16:00:00Z through 2026-10-01T20:00:00Z; validate continuity before execution. This is a falsifiable relative-liquidity hypothesis, not an assumption of permanent cointegration. No funding income is included.

Hyperliquid represents 6.8% of experiments against a 20% target. Two opposing positions address the 84% long-only and 65.6% single-symbol concentrations. Corpus concentration was empty, so instrument-specific research shares cannot be established. Unlike the failed same-coin cross-exchange arbitrage, this tests multi-percentage-point cross-asset shocks over days, with explicit exhaustion confirmation and no fitted OU model. Options were rejected because the supplied outcomes identify structural cost failures. The provided library excerpts do not substantiate this mechanism.

verification_loop: Verification failed (Layer 1.5 — hypothesis/config consistency) [class=hypothesis_mismatch]: - optimization_plan fixes leverage at 1.0

Implementation

Trades opposing, approximately equal-dollar LTC and BCH Hyperliquid perpetual positions after a six-bar relative shock and observable partial reversal. Uses synchronized four-hour closes, rolling return correlation and shock volatility, a relative-spread stop, reversal target, 18-bar holding limit, six-bar cooldown and catastrophic per-leg close-based stops.

Preserves the previous continuous signal, bounded rolling covariance, exact instruments, synchronization and research parameter permissions. The timeout feedback contains no profiling evidence, and the existing indicator work is already constant-time. This focused iteration addresses a possible synchronous event storm: pair-close dispatch is guarded against re-entry, position-close callbacks no longer cancel or duplicate an active pair close, and rejected closes are retried on a subsequent decision bar instead of recursively inside rejection callbacks. A synchronous primary-entry rejection also prevents subsequent hedge submission. No verification or final-exam data reads were performed; timeout resolution remains for the pipeline to establish. If the timeout persists, engineering should profile catalog loading and engine execution rather than infer alpha failure. The required outbox file could not be written because this session's filesystem is read-only; the complete call-bound JSON is returned here.

Verification Results

Verification failed (Layer 1.5 — hypothesis/config consistency) [class=hypothesis_mismatch]: - optimization_plan fixes leverage at 1.0

Iteration History

sandbox_timeout: Verification failed (Layer 3 — sandbox backtest): smoke test exceeded the 300s wall-clock limit on a ~20000-bar window (2400 days of LTCUSD.HYPERLIQUID-4-HOUR-LAST-EXTERNAL). This almost always means per-bar work that scales with history — e.g. rescanning the full funding/supplementary series, or rebuilding a list and calling min()/sorted() inside calculate_signal()/on_bar() on every bar. Precompute sorted timestamp arrays ONCE in __init__ and use bisect, or cache lookups keyed by timestamp, so per-bar cost is O(log n) not O(n).

Outcome Summary

Make sure the strategy config's leverage agrees with the hypothesis's optimization_plan before resubmitting, because a repeated consistency mismatch ends the run before any backtest happens.

After 3 iterations it was abandoned in the verification loop. It failed Layer 1.5 (hypothesis/config consistency, class=hypothesis_mismatch) because the optimization_plan fixes leverage at 1.0.

It tested whether unusually large relative moves between LTC and BCH perpetuals on Hyperliquid partly reverse after visible exhaustion. It traded equal-dollar, opposite positions on 4-hour bars and did not assume permanent cointegration or count any funding income.

No performance figures exist: the backtest report is empty and optimization never ran, so there are no return, Sharpe or trade-count numbers.