SuiFourHourVolumeConfirmedDonchianBreakoutLong

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

Long-only SUI 4-hour strategy on BINANCE USD-M futures. Direct analog of the landed XRP (Sharpe 1.84, paper_stage) / BTC / ETH / SOL / BNB / AVAX / OP / INJ / TIA FourHourVolumeBreakoutLong family — applied to SUI. The 2.0x volume surge trigger produces ~30-60 trades/year (statistically valid for walk-forward), explicitly avoiding the low-frequency channel-breakout pathology that killed ADA daily. Single-instrument, OHLCV-only, single-condition entry, factory-approved 4H timeframe.

Iteration 2 implements the analyst's exact remedy for the optimization-destroyed base config. The prior run searched ~6-7 parameters and over-selected the 3 cliff params (volume_multiplier driven to 2.93, plus stop_atr_multiple and volume_period) into a 16-trade config with a degenerate zero-trade holdout (DSR 0.2447, PBO 0.7765, 3 sensitivity cliffs), deflating a clean base config (Sharpe 1.31 on 58 trades). Because the strategy reads every parameter via self.parameters.get(key, family_default), I shrink the SEARCH SURFACE purely through the config: the parameters dict now exposes ONLY breakout_period, so volume_multiplier=2.0, stop_atr_multiple=2.0, volume_period=20, atr_period=14, target_atr_multiple=4.0, and capital_fraction=0.2 are all fixed at their XRP-validated family defaults via the code's fallbacks and are removed from optimization. This is the softer alternative to hard-coding constants (params remain overridable, so sensitivity/walk-forward still have a live surface on breakout_period and the robustness gate is NOT structurally defeated), which restores the ~58-trade base config and should yield DSR >= 0.95 with a non-degenerate holdout. No code/signal/import changes were made — only the exposed search surface — so the Layer 1-3 passes are preserved. Per the analyst's abandon condition, if this family-standard re-run still produces a zero-trade holdout, SUI's ~3-year history is simply too short to validate a selective 4H volume breakout via walk-forward and the strategy should be abandoned (the family's edge is already captured by the longer-history XRP/BTC/ETH/SOL siblings).

Failed deflated Sharpe and a negative forward holdout on attempt 2 of 2. DSR=0.409 (vs 0.95 bar) with the optimized Sharpe 1.26 BELOW the expected-max best-of-225 luck bar of 1.4497 (is_significant=false, PBO 0.705, sharpe_ci_low -0.61 straddling 0) — after multiple-testing correction the selected config is statistically indistinguishable from best-of-N noise. Decisively, the time-ordered HOLDOUT FAILED and is NEGATIVE (holdout_sharpe -1.757, ratio -0.84 vs WF-OOS 2.09): the untouched recent window loses money, reflecting genuine regime decay (annual return 2024 +21% -> 2025 +2.2% -> 2026 -1.6%; rolling Sharpe negative by mid-2026). The PSR=0.933 / DSR=0.409 split is the diagnostic tell of 225-trial selection inflation; a clean sensitivity grid (0 cliffs) and a non-overfit walk-forward (avg OOS 2.09) measure consistency, not significance, and do not override the deflation failure plus a negative holdout. Not iterate: this is attempt 2 of 2 and iteration 2 already applied the prescribed fix (search surface shrunk to only breakout_period, family params locked) which did not rescue the holdout or DSR — there is no further tunable region. Not revise_hypothesis: the volume-confirmed-breakout mechanism is already validated/promoted on many siblings (XRP 1.84, plus BTC/ETH/SOL/BNB/AVAX/OP/INJ/TIA), so this is SUI-specific asset-selectivity, not a sound mechanism stranded on a dead target. FAILURE PATTERN: the FourHourVolumeBreakoutLong family does NOT transfer to every major-alt arm — on SUI (short ~3yr history, ~60 trades) the edge is too modest to clear best-of-225 deflation (optimized Sharpe 1.26 below the 1.45 luck bar, DSR 0.409, PBO 0.705) and the recent untouched holdout is NEGATIVE (-1.76) due to 2025-2026 decay. A validated family and a non-overfit walk-forward do not rescue an arm that fails deflated Sharpe and a negative forward holdout; the promoted siblings are the assets where this mechanism clears the bar, SUI is not.

Long-only SUI 4H volume-confirmed Donchian breakout on BINANCE USD-M futures, a direct analog of the validated XRP/BTC/ETH/SOL FourHourVolumeBreakoutLong family. Enters long when the bar breaks above the prior-20-bar Donchian high AND volume >= 2.0x the 20-bar average; exits on an ATR bracket (2.0x ATR stop / 4.0x ATR target). Only breakout_period is exposed to optimization; all other params are fixed at family-standard defaults.

Direct analog of a landed/validated family (XRP FourHourVolumeBreakoutLong at paper_stage Sharpe 1.84, plus 8+ siblings) — the mechanism class is proven

Functioning and well-behaved: profit_factor 1.372 (>1.2), Sharpe 1.163, 60 trades (~20/yr, walk-forward-adequate), 0 entries dropped

NOT outlier-carried: return_kurtosis 8.4 (low/moderate), tail_ratio 1.61, healthy positive skew — clean return distribution

Strong risk profile: max_drawdown 10.1%, exposure 18.8%; positive across 2023/2024/2025 with only a mild 2026-YTD dip

Iteration-2 fix is correct: search surface shrunk to breakout_period only, cliff params fixed at XRP-validated defaults — directly addresses the prior over-selection failure

Base Sharpe 1.163 not yet significant: sharpe_ci_low -0.748 (CI straddles 0), PSR 0.896 — real risk of failing the 225-trial deflated-Sharpe gate after optimization

Shorter history (SUI listed May 2023, ~3yr) — fewer regimes tested than the 6-year majors

2026 YTD negative (-4.9%) — worth watching for recent-regime decay in the post-opt holdout

Member of a validated family (XRP sibling Sharpe 1.84, paper_stage) with clean structure: single-instrument, OHLCV-only, single-condition entry

Sensitivity is clean (0 cliffs on breakout_period) and walk-forward is not overfit (avg OOS 2.09, is_overfitted=false)

Healthy trade economics in-sample: PF 1.41, avg_position 20%, low impact_cost 2.2%, large capacity ($210M) — costs are not the problem

Failed deflated Sharpe: DSR 0.409 << 0.95, is_significant=false; optimized Sharpe 1.26 is BELOW the 225-trial expected-max luck bar of 1.45 — indistinguishable from best-of-N noise

Holdout FAILED and is NEGATIVE: holdout_sharpe -1.757 (ratio -0.84) vs WF-OOS 2.09 — the untouched recent window loses money

PBO 0.705 (>0.5): parameter selection is more likely than not overfit

sharpe_ci_low -0.61 straddles 0; probabilistic_sharpe 0.933 < 0.95

Regime decay: annual returns fall 2024 +21% -> 2025 +2.2% -> 2026 -1.6%, with rolling Sharpe turning negative by mid-2026 — exactly what the negative holdout catches

Short history / thin sample: ~3 years, only 60-68 trades total, so best-of-225 selection over a modest edge inflates easily

Optimization DESTROYED a clean, family-validated base config rather than improving it — fix by shrinking the search back to the validated family-standard parameters. The base config (breakout 20, vol_mult 2.0, atr 14, vol_period 20, stop 2.0, target 4.0) gave Sharpe 1.31 on 58 trades; the optimizer over-selected to a 16-trade config (volume_multiplier driven to 2.93, parked on a cliff) that deflates (DSR 0.2447, PBO 0.7765), fails sensitivity (3 cliffs: volume_multiplier, stop_atr_multiple, volume_period), and produces a DEGENERATE holdout (holdout_sharpe 0.0 — no trades in the final window). The edge is real: the family is validated (XRP 1.84 + 8 siblings) and 2 of 3 walk-forward OOS windows are strongly positive [3.30, 3.04, 0.0] (the 3rd is empty from SUI's short ~3yr history, not negative). FIX: drastically shrink the parameter surface to the XRP-validated family-standard values — FIX the 3 cliff parameters at their family defaults (volume_multiplier=2.0, stop_atr_multiple=2.0, volume_period=20) and fix the other structural params (atr_period=14, target_atr_multiple=4.0, capital_fraction=0.2) at family defaults; optimize at most breakout_period in a narrow band (18-22), or hard-lock ALL params to the family standard and run the validation pass (as was done for the ADA volume-absorption resubmission). This restores the ~58-trade base config and should yield a NON-degenerate holdout. TARGET: DSR >= 0.95 AND a holdout with actual trades (holdout_sharpe > 0, ratio >= 0.70) AND sensitivity passing. CRITICAL ABANDON CONDITION: if the family-standard re-run STILL produces a degenerate (zero-trade) holdout, SUI's ~3-year history is simply too short to validate a selective 4H volume-breakout via walk-forward — abandon then (a data-length limitation, not a tuning problem), since the family's value is already captured by the longer-history siblings (XRP/BTC/ETH/SOL).

This strategy ported the promoted FourHourVolumeBreakoutLong mechanism — whose XRP sibling reached paper stage at Sharpe 1.84 — onto SUI, and its iteration-2 fix correctly shrank the search surface to just breakout_period while locking the family-standard params. It passed the pre-optimization review (verdict: optimize) as a clean, well-behaved member of a proven family with Sharpe 1.163 and a 10.1% drawdown, though its confidence interval straddled zero. Optimization confirmed the flagged risk: deflated Sharpe 0.409, an optimized Sharpe below the 1.45 luck bar, PBO 0.705, and a negative untouched holdout as annual returns decayed from +21% (2024) to -1.6% (2026). The analyst concluded this was SUI-specific asset-selectivity rather than a mechanism or tuning flaw — the promoted siblings remain the assets where the edge clears the bar — and abandoned it after exhausting both attempts.

A validated strategy family does not transfer to every asset arm: on SUI's short ~3-year history and modest ~60-trade sample the edge was too thin to clear best-of-225 deflation, and a clean sensitivity grid plus a non-overfit walk-forward measure consistency, not significance — they cannot override a deflation failure and a negative forward holdout.

The analyst abandoned it at the post-optimization ANALYZING stage (iteration 2 of 2): it failed the deflated Sharpe test (DSR 0.409 vs 0.95) with the optimized Sharpe 1.257 below the 225-trial expected-max luck bar of 1.45, PBO 0.705, and — decisively — a FAILED, negative time-ordered holdout (holdout Sharpe -1.757, ratio -0.84 vs WF-OOS 2.09) reflecting 2025-2026 regime decay.

A long-only SUIUSDT 4H volume-confirmed Donchian breakout on Binance USD-M futures — entering when a bar breaks the prior N-bar Donchian high on a 2.0x volume surge, exiting via an ATR bracket — as a direct analog of the validated XRP/BTC/ETH/SOL FourHourVolumeBreakoutLong family applied to SUI.

The base backtest showed Sharpe 1.163, +24.3% total return over 60 trades, profit factor 1.372, and a low 10.1% max drawdown; optimization (breakout_period 12) lifted it to Sharpe 1.257, +32% return, 68 trades, and profit factor 1.414, but with deflated Sharpe of only 0.409 and a negative holdout.