LTC Bybit R-Breaker: Prior-Day Breakout and Failed-Extension Reversal

Hypothesis

Test the source's combined breakout and reversal mechanism on LTCUSDT.BYBIT using 1-MINUTE bars, with positions held for hours rather than minute scalps. Freeze six reference levels from the preceding complete UTC day's high H, low L and final close C. Use the prose's previous-day interpretation rather than the sample's rolling two-record extrema and minimum close. Maintain daily extrema and observation flags incrementally. Implement as a FactoryStrategy subclass with bounded state and constant-time per-bar work.

This preserves the submitted R-Breaker mechanism while adding explicit cost and execution controls. Bybit is underrepresented at 2.3%, and both long and short trades address the portfolio's 83.2% long-only concentration. Hyperliquid's minute-history constraint makes it unsuitable for this test despite its larger venue deficit. The single-instrument scope follows the user's mechanism. Corpus concentration was empty, so LTC-specific experiment share is unknown. Negative-Sharpe execution checks are not evidence of an economic edge.

negative_expectancy: decline optimization of this LTC Bybit prior-day R-Breaker implementation. Executed economics lose substantially across 916 closed positions, including after removing modeled impact. No verified implementation repair supports developer iteration. This rejects the tested specification, not R-Breaker generally or every alternative parameter configuration. The requested outbox could not be written because this session permits read-only filesystem access.

Implementation

R-Breaker on LTCUSDT.BYBIT 1-minute bars, long and short. Six levels are frozen from the previous complete UTC day's high, low and close. It trades breakouts beyond the break levels. It also trades failed extensions: price touches a setup level, then crosses back through the matching enter level. Each entry gets a venue bracket anchored at entry: stop at 0.2x the previous day's range, target at the larger of 0.45x that range or 1.2%. Size targets 0.35% of equity at risk, capped at 30% notional. Limits: at most 2 entries a day, 120-minute cooldown after each exit, 720-minute maximum hold, all positions flattened at 23:55 UTC, 1% daily loss limit.

Iteration 2 fixes the Layer-1.5 hypothesis_mismatch. The parameters now contain every fixed value from optimization_plan with its exact type, including leverage=1.0, which was missing. They also carry the tunable starting values f1=0.45 and f3=0.25, both inside their declared ranges. The plan passes a local check with src/optimization/domains.validate_domains. The code change is small: the strategy reads the `leverage` parameter, and the exposure cap uses the lower of it and config.leverage. Signal and entry logic are unchanged.

Verification Results

The truncated sandbox closed seven losing positions and reported substantial modeled impact costs. Reconcile impact assumptions and trade-level economics in the full backtest; these smoke-test losses do not establish a correctness defect.

Backtest Review

Sharpe

-2.48

Total return

-100.00%

Max drawdown

100.00%

Trades

916

Win rate

31.9%

Profit factor

0.27

916 closed positions provide a substantial executed sample; balanced directions and 2h 11m average holding are compatible with the stated mechanism.

Net closed-position PnL totals -157,068.70 USDT. Adding back all 126,643.02 USDT of modeled impact still leaves -30,425.68 USDT; impact alone does not explain the losses. QA's economic concern was borne out.

PF is 0.271; average trade return is -0.854% versus the +0.15% futures review floor. Average losses of 345.42 USDT exceed average wins of 200.26 USDT.

Reported liquidation in October 2023 conflicts with trades continuing through September 2026. Reconcile impact-adjusted equity, sizing and liquidation before reusing headline risk metrics; the discrepancy's mechanism is unverified.

Buy-and-hold returned -8.34% with approximately 87.94% drawdown on the detail curve. Exact entry-condition fidelity cannot be verified from trade records alone.

Iteration History

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

Outcome Summary

Any future iteration should reconcile equity, sizing and liquidation accounting and establish positive executed economics before optimization.

After two iterations, backtest review returned abandon for negative expectancy, with no verified implementation repair supporting another iteration. Optimization, post-optimization analysis and risk review were not reached.

Trade LTCUSDT on Bybit using prior-day R-Breaker levels for breakouts and failed-extension reversals on one-minute bars, holding positions for hours.

Across 916 closed positions, the backtest reported -100% return, Sharpe -2.48, profit factor 0.271 and average trade return -0.854%. The review found losses remained substantial after removing modeled impact and flagged reported liquidation as inconsistent with subsequent trades.

Analysis

R-Breaker trading strategy