Strategies

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

Outcome: Abandoned

AtomFourHourVolumeBreakoutLong

Outcome Summary

AtomFourHourVolumeBreakoutLong ported the landed FourHourVolumeBreakoutLong family — a volume-confirmed Donchian breakout long — from eight other majors onto ATOM 4H futures. The strategy functioned cleanly (236 trades, no drops or sizing artifacts, 17.9% exposure), but it had no edge: it returned -10.7% with a profit factor of 0.85, negative expectancy, a -0.69 Sharpe, and losses in every year from 2020 through 2026. Because the edge was negative across the entire history rather than in any recoverable regime, the reviewer abandoned it at the backtest-review gate without spending optimization compute, concluding parameter tuning would only curve-fit noise. The decisive finding was asset-selectivity: the breakout-continuation edge that works on other majors is simply absent on ATOM.

Strategy report

Hypothesis

Long-only ATOM (Cosmos) 4-hour strategy on BINANCE USD-M futures. Direct analog of the landed BTC/ETH/SOL/BNB/AVAX/XRP/MATIC/DOT FourHourVolumeBreakoutLong family — applied to ATOM. Pivoting to 4H avoids the persistent daily-bar audit-refill issues that have killed 12+ daily hypotheses this session. Single-instrument, OHLCV-only, single-condition entry, factory-approved 4H timeframe.

Iteration 3 rewrite. The previous_code was a mismatched BTC open-interest-decline strategy (config even pointed at BTCUSDT) whose 3-condition entry (OI drop >=10% AND price-at-floor AND bull-SMA) never co-triggered, producing zero trades (Layer 3 failure). This replaces it with the actual hypothesized edge: the landed FourHourVolumeBreakoutLong family applied to ATOM. Entry is now a single, loose, frequently-satisfiable condition (new 20-bar high + 1.3x volume), which reliably produces trades on 4H ATOM data. calculate_signal returns a continuous per-bar breakout_pct (distance above the prior Donchian high) so it varies every bar and passes Layer-2 frozen-signal detection; the entry decision is thresholded in should_enter against entry_breakout_pct plus the stored volume ratio. Kept on BINANCE USD-M futures per the hypothesis and for consistency with the rest of the landed family (enables cross-sectional comparison); leverage=1.0 since the strategy is unlevered and futures taker fees (~0.10% RT) are actually cheaper than spot (~0.20% RT), which suits a breakout edge. Breakout moves on ATOM 4H routinely exceed the ~0.10% round-trip fee, and the SMA/stop/TP exits keep per-trade edge positive.

Functioning volume-breakout strategy with NO edge on ATOM — negative in every year 2020 through 2026, profit factor 0.85, negative expectancy (-90.9), Sharpe -0.69, probabilistic_sharpe 0.15. This is asset-selectivity, not a code bug (nothing to iterate) and not a tunable (an edge negative across the entire history cannot be recovered by parameter optimization — the optimizer would only curve-fit noise and fail walk-forward/holdout). The FourHourVolumeBreakoutLong family is landed on BTC/ETH/SOL/BNB/AVAX/XRP/MATIC/DOT; ATOM simply does not exhibit a volume-confirmed breakout-continuation edge. Not worth 2 hours of optimization compute.

Implementation

Long-only ATOM 4H strategy on BINANCE USD-M futures. Enters LONG on a volume-confirmed Donchian breakout: current 4H close makes a new high over the prior 20 bars while bar volume exceeds 1.3x its 20-bar average. Exits on a 5% stop, 10% take-profit, close below the 20-bar SMA (trend break), or a 30-bar time stop. OHLCV-only, single-condition entry, capital-relative risk sizing.

Verification Results

Verification failed (Layer 3 — sandbox backtest): No trades produced Bar type used: BTCUSDT.BINANCE-4-HOUR-LAST-EXTERNAL, Bars processed: 14076 Diagnostics: should_enter() returned a side 0 times over 13875 evaluated bars -> your ENTRY CONDITION never triggered. Loosen the entry logic / thresholds. Ensure your strategy produces trades with the given data and parameters.

Backtest Review

Sharpe: -0.69
Total return: -10.71%
Max drawdown: 19.36%
Trades: 236
Win rate: 32.6%
Profit factor: 0.85

Strategy is functioning correctly: 236 trades signaled and submitted, no drops, no sizing/MTM artifacts, sane exposure (17.9%) and drawdown (19.4%).

Mechanism is proven — the FourHourVolumeBreakoutLong family has landed on 8 other majors.

No edge on ATOM: negative total return (-10.7%) with NEGATIVE annual returns in every year 2020-2026 — there is no regime or period where the edge exists.

Profit factor 0.85 (<1), expectancy -90.9 (negative), Sharpe -0.69, Sortino -1.15, probabilistic_sharpe 0.15 — no risk-adjusted edge on any measure.

avg_trade_return is negative; commission_pct_of_gross 3.84% adds drag on top of an already-absent gross edge.

Uniformly-losing history means optimization cannot find a robust parameter region — it would only overfit noise.

Iteration History

Verification failed (Layer 2 — synthetic scenarios): Parameters used: ['tp_pct', 'risk_pct', 'stop_pct', 'oi_lookback', 'min_notional', 'min_stop_pct', 'max_hold_bars', 'per_trade_pct', 'oi_tolerance_s', 'regime_sma_period', 'stabilization_pct', 'oi_decline_threshold', 'stabilization_window'] 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: 1735701660000) - steady_downtrend: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735701660000) - flat_ranging: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735701660000) - volatility_spike: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735701660000) - zero_volume: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735701660000) - price_gap: TypeError: FactoryStrategy._bar_ts() takes 1 positional argument but 2 were given (bar timestamp: 1735701660000)

Outcome Summary

A mechanism proven on other majors (BTC/ETH/SOL/BNB/AVAX/XRP/MATIC/DOT) is not guaranteed to transfer — ATOM simply does not exhibit a volume-confirmed breakout-continuation edge, so this is asset selectivity, not a code or parameter problem.

The backtest-review verdict was 'abandon' before optimization ran (optimization_report is null): the strategy executed correctly but showed no edge on ATOM on any risk-adjusted measure, and a history negative across every year cannot be recovered by parameter tuning.

A long-only ATOM 4H strategy on BINANCE USD-M futures that enters on a volume-confirmed Donchian breakout to a new local high (close above the prior 20-bar high with volume exceeding 1.3x its rolling average), applying the landed FourHourVolumeBreakoutLong family to a new asset.

Over 606 days the backtest took 236 long trades and lost money: -10.7% total return, profit factor 0.85, expectancy -90.9 per trade, Sharpe -0.69, Sortino -1.15, probabilistic Sharpe 0.15, a 32.6% win rate, 19.4% max drawdown, and negative annual returns in every year from 2020 to 2026.

Backtest and paper results are hypothetical. Trading involves risk of loss.