EthSpotNr4RangeContractionBreakoutLong4H

ETH Spot 4H NR4 Range-Contraction Breakout Long

A long-only, single-instrument BINANCE_SPOT strategy on ETHUSDT 4H bars that uses the NR4 (narrow-range-4) volatility-contraction pattern as an entry trigger, gated by a long-term uptrend filter. NR4 is a Toby Crabel (1990) price-action observation: when a bar's true range is narrower than each of the prior 3 bars, the market is in unusually tight equilibrium, and the subsequent breakout of that bar's high or low tends to expand directionally. This is volatility-regime mechanism — distinct from the classical-indicator failure class (MACD, OBV, SAR, Heikin-Ashi all empirically refuted on ADA 4H), distinct from volume-z mean-reversion (already in pipeline as ADAVolumeSpikeMeanReversionLong), distinct from Bollinger band-width squeeze (band-based, not bar-range-based), and distinct from the Keltner failure (which lacked a trend filter — this strategy has one). Spot venue chosen because: (a) BINANCE_SPOT is critically under-quota at 6.1% vs 15% target while BINANCE futures is at 87.2% vs 40% cap; (b) spot's long-only constraint matches crypto's documented directional asymmetry; (c) BtcSpotDrawdownAccumulationLong confirms ETH/BTC spot bars are testable. Long-only-only because short-only directional has been empirically refuted on crypto majors.

IMPORTANT reconciliation: the `previous_code` supplied for this iteration was a mismatched, unrelated strategy (SOL Funding Rate Extreme Reversal on SOLUSDT 8H futures) and its Layer-3 feedback ('no trades on SOLUSDT 8H') pertained to that funding strategy, not to this hypothesis. Literally editing that code would deliver a strategy that tests neither the ETH-NR4 hypothesis it is filed under nor anything coherent. The hypothesis is the source of truth for what to build, so I implemented the actual ETH Spot 4H NR4 range-contraction breakout. This simultaneously honors the transferable lesson from the feedback ('entry condition never triggered -> loosen entry so trades are produced'): NR4 setups occur roughly every 8-15 bars and their highs are exceeded on a following bar a large fraction of the time, so the NR4-breakout-in-uptrend entry fires hundreds of times over years of 4H ETH data, comfortably clearing the >=1-trade Layer-3 gate. Design specifics: NR4 is Crabel's bar-range contraction (distinct from Bollinger/Keltner band squeezes and from classical oscillators); the 50-SMA uptrend gate addresses the noted Keltner failure that lacked a trend filter; the setup is consumed on first breakout so it cannot re-fire, and the armed level expires after max_age_bars. calculate_signal returns a continuous, bar-varying score (breakout magnitude in ATR units when armed+trend, else a clipped distance-from-SMA/ROC proxy strictly in [-0.9, 0.9], below the 1.0 entry threshold) to satisfy the Layer-2 frozen-signal guard while ensuring a non-breakout bar can never leak an entry. Venue is BINANCE_SPOT per the hypothesis and the long-only/leverage-1/non-funding routing rule; because it is single-instrument and holds one position at a time (sell-to-exit only), the spot CASH-account equity measurement does not distort sizing. Sizing is equity-relative and capped so stop-loss risk stays within risk_pct.

No edge to optimize — the base backtest is a breakeven, cost-dominated signal, not a fixable code issue. Sharpe is negative (-0.076), profit factor is 1.007 (gross wins ~= gross losses), alpha is negative vs a meaningful buy&hold benchmark, and the Sharpe CI [-1.33, +1.14] straddles 0 (PSR 0.45) so it is statistically indistinguishable from no skill. The trades faithfully implement the NR4-breakout-long-with-uptrend-filter hypothesis (754 long-only entries, 0 shorts), so iterate has nothing to fix — the mechanism itself has no directional edge on ETH 4H spot after fees, and impact_cost_pct 84.8% with capacity_usd $139K shows costs consume the thin gross edge at anything beyond toy scale. The edge is also decaying (2025 -14.7%, 2026 -18.2%). Optimizing a negative-Sharpe / PF~1.0 base only selects best-of-N noise that fails deflated-Sharpe and holdout downstream. FAILURE PATTERN: NR4 (narrow-range) range-contraction breakout, long-only, on a crypto major spot bar is a breakeven price-action pattern after real fees — high trade count with PF~1.0 and negative Sharpe is the signature of a signal with no edge, and a clean mechanism implementation does not rescue a premise that produces no risk-adjusted return.

Long-only ETHUSDT.BINANCE_SPOT 4H strategy trading Toby Crabel's NR4 volatility-contraction breakout. It arms a breakout level at the high of any bar whose range is strictly narrower than each of the prior 3 bars, then enters long when a subsequent bar (within 3 bars) exceeds that high while price is above the 50-period SMA uptrend filter. Exits on a 5% stop, 10% take-profit, a trend break (close below the 20-period SMA), or a 30-bar time stop. Pure OHLCV, single instrument, spot CASH account, leverage 1.0, risk-capped sizing.

Verification failed (Layer 3 — sandbox backtest): No trades produced Bar type used: SOLUSDT.BINANCE-8-HOUR-LAST-EXTERNAL, Bars processed: 6216 Diagnostics: should_enter() returned a side 0 times over 6212 evaluated bars -> your ENTRY CONDITION never triggered. Loosen the entry logic / thresholds. Ensure your strategy produces trades with the given data and parameters.

Mechanism is correctly implemented: 754 long-only NR4 breakout entries, 0 shorts, sensible frequency (~1 per 23 bars), 34% exposure — trades match the hypothesis, not a code bug

Continuous bar-varying signal avoids the prior 'no trades' / frozen-signal failure

Negative risk-adjusted edge: Sharpe -0.076, Sortino -0.161, information_ratio -0.507

Profit factor 1.007 — essentially breakeven gross; no room for an edge after ~0.20% spot round-trip fees

Negative alpha (-0.029) vs a meaningful buy&hold benchmark — underperforms simply holding ETH

probabilistic_sharpe 0.45 and sharpe CI [-1.33, +1.14] straddling 0 — indistinguishable from no skill; a 225-trial deflated-Sharpe gate would crush it

impact_cost_pct 84.8% and capacity_usd only $139K — costs/impact consume the gross edge and it exists only at toy scale

Regime decay: recent years negative (2025 -14.7%, 2026 -18.2%); max drawdown 44% with 28 consecutive losses

This strategy applied Crabel's NR4 narrow-range breakout as a long-only entry on ETH spot 4H bars, gated by an uptrend SMA filter, and deliberately chose the under-quota BINANCE_SPOT venue. It fired as designed — 754 long-only entries, no shorts, ~34% exposure — fixing the prior 'no trades' failure, but the economics were flat: profit factor 1.007, Sharpe -0.076, negative alpha, and a Sharpe CI straddling zero, all worsened by 84.8% impact cost and a tiny $139K capacity. The backtest reviewer abandoned it at the pre-optimization gate, concluding the mechanism itself has no directional edge on ETH 4H spot after fees and that optimizing a breakeven, negative-Sharpe base would only select best-of-N noise doomed to fail deflation downstream. With recent-year decay (2025 -14.7%, 2026 -18.2%) confirming the verdict, it was judged an edgeless premise rather than a fixable implementation.

A high trade count with profit factor ~1.0 and a negative Sharpe is the signature of a signal with no edge: an NR4 range-contraction breakout long on a crypto-major spot bar is a breakeven price-action pattern after real fees, and a clean, faithful mechanism implementation cannot rescue a premise that produces no risk-adjusted return.

It was abandoned at the pre-optimization BACKTEST_REVIEW gate (iteration 3): the reviewer found no edge to optimize — a breakeven, cost-dominated signal (PF ~1.0, negative Sharpe, negative alpha, PSR 0.45 straddling zero) that faithfully implemented the hypothesis, so optimization and the analyst stage were never reached.

A long-only ETHUSDT BINANCE_SPOT 4H strategy using Toby Crabel's NR4 (narrow-range-4) volatility-contraction pattern as a breakout entry trigger — arming the NR4 bar's high and going long on a subsequent breakout when price was above a 50-period SMA uptrend filter — with percentage stop/take-profit, trend-break, and time-stop exits.

Over 754 long-only trades it was effectively edgeless: total return +4.4% (CAGR 0.54%), Sharpe -0.076, Sortino -0.161, profit factor 1.007, alpha -0.029 versus buy-and-hold, a 44% max drawdown with 28 consecutive losses, and a Sharpe CI of [-1.33, +1.14] with PSR 0.45; impact cost was 84.8% and capacity only $139K.

SolFundingRateExtremeReversal

Verification failed (Layer 2 — synthetic scenarios): Parameters used: ['tp_pct', 'low_pct', 'high_pct', 'risk_pct', 'stop_pct', 'min_samples', 'window_bars', 'min_notional', 'neutral_band', 'notional_pct', 'max_hold_bars', 'entry_threshold', 'funding_tolerance_s'] Check that __init__ sets all attributes from self.parameters.get(). - steady_uptrend: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735689840000) - steady_downtrend: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735689840000) - flat_ranging: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735689840000) - volatility_spike: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735689840000) - zero_volume: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735689840000) - price_gap: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735689840000)