ArbFourHourVolumeBreakoutLong4H

ARB Four-Hour Volume Breakout Long-Only (BINANCE USD-M Futures, 4H, OHLCV-Only)

Long-only ARB (Arbitrum) 4-hour strategy on BINANCE USD-M futures. Direct analog of the landed BTC/ETH/SOL/BNB/AVAX/XRP/APT/FIL/ICP FourHourVolumeBreakoutLong family — applied to ARB. Pivoting to 4H avoids the persistent daily-bar audit-refill issues. Single-instrument, OHLCV-only, single-condition entry, factory-approved 4H timeframe. The 2.0x volume surge trigger (not channel breakout) keeps trade count high enough (~30-60/year) to be statistically meaningful for walk-forward optimization — explicitly avoiding the low-frequency channel-breakout pathology that killed the ADA daily multi-week trend strategy.

Direct analog of the landed BTC/ETH/SOL/BNB/AVAX/XRP/APT/FIL/ICP FourHourVolumeBreakoutLong family, applied to ARB exactly as the hypothesis requests. Uses the proven 4H timeframe (avoiding daily-bar audit-refill issues) and the 2.0x volume-surge trigger (rather than a low-frequency channel breakout) to keep trade count in the ~30-60/year range needed for meaningful walk-forward optimization, explicitly sidestepping the low-frequency pathology that killed the ADA daily channel-breakout strategy. Single-instrument, OHLCV-only, single-condition core entry (volume surge) with standard trend + bullish-bar filters. BINANCE USD-M futures venue keeps round-trip taker fees at ~0.10% (cheaper than spot), and the +8%/-4% exit asymmetry plus multi-bar holds ensures average per-trade returns comfortably clear trading costs. Volume mean excludes the current bar to prevent self-comparison bias; ARB qty precision (0.1 step) is respected in position_size.

No robust edge to optimize. Base Sharpe is 0.464 with a bootstrap CI [-0.357, 1.285] that straddles zero, information_ratio is only 0.215, and the +22% total return is outlier-driven (kurtosis 20.5) and concentrated in 2026 (+15.7%) while 2025 lost -5.1% with rolling Sharpe going deeply negative. This is a single-name, OHLCV-only, single-condition directional volume-breakout — the trend/breakout-clone family with a near-zero-survivor track record (L9/L41, ~0/213), and the APT sibling of this exact 'landed' FourHourVolumeBreakoutLong family was already abandoned at backtest_review as negative-expectancy, so the family framing does not hold. It is not a fee failure (avg_trade_return_pct 1.12%), but a sub-1.0 Sharpe on this class cannot be lifted to a promotable, deflation-surviving level by tuning 6 parameters — the optimizer would curve-fit the few outlier winners and die at deflated-Sharpe/holdout. Not worth 2 hours of optimization.

Long-only ARB (Arbitrum) 4H volume-surge breakout on BINANCE USD-M futures, OHLCV-only. calculate_signal returns a continuous volume-surge score (tanh of the excess of the current bar's volume ratio over the prior-window mean), varying every bar. Entry requires three conditions on the just-closed 4H bar: (1) volume >= 2.0x the mean of the prior 20 bars' volumes (current bar excluded to avoid self-comparison bias), (2) bullish bar (close > open), and (3) close above the 50-bar SMA uptrend filter. Exits on take-profit (+8%), stop-loss (-4%), or a 30-bar time stop. Sizing is equity-relative at 20% notional per position, no leverage.

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.

avg_trade_return_pct 1.12% clears the Binance USD-M fee floor comfortably (not fee-limited)

profit_factor 1.28 is above the 1.2 line and max_drawdown 11.9% is well controlled

110 trades over 3.4 years (~32/yr) is an adequate sample for walk-forward

entry diagnostics clean (110 signaled, 110 submitted, no drops); positive alpha 0.050, low beta

Base Sharpe 0.464 with CI [-0.357, 1.285] straddling zero — no statistically significant risk-adjusted edge

Single-name OHLCV directional volume-breakout — the trend/breakout-clone family has ~0/213 survivors (L9/L41)

Return is outlier-driven (kurtosis 20.5, skew 1.77) and concentrated in 2026 (+15.7%); 2025 was -5.1% with rolling Sharpe deeply negative

The 'landed family' claim is unreliable — the APT sibling of this exact FourHourVolumeBreakoutLong family was abandoned at backtest_review as negative-expectancy

Sharpe < 1.0 on this class cannot reach a promotable, deflation-surviving level through parameter tuning

data_unavailable: backtest data for ARBUSDT.BINANCE could not be collected after 5 retries: - bars (bar_type=ARBUSDT.BINANCE-4-HOUR-LAST-EXTERNAL): got 0, needed 100, or 60+ days span The data layer (not the strategy) is the blocker — re-evaluate this hypothesis once collection is fixed / the symbol is backfilled.