AtomBinanceDualTimeframeMomentumConfluenceLS

ATOM Binance USD-M Dual-Timeframe Momentum Confluence — Long-Short 4H Trend Aligned with the Daily Trend, Flat on Disagreement, ATR Trailing Exit (Volatility-Scaled, Single-Bar, 3-Parameter)

A LONG-SHORT, single-instrument, pure-price MOMENTUM-CONFLUENCE trend follower on ATOMUSDT.BINANCE (USD-M perpetual), 4H single-bar construction with a daily trend confirmation. This is an EVIDENCE-DRIVEN asset selection following the session's single most informative result: the ADA dual-TF confluence NEAR-MISS, which was 'the strongest dual-TF candidate the factory has produced and the ONLY one to emphatically clear significance' (deflated_sharpe 1.0 vs expected-max 0.56, optimized Sharpe 1.75, positive CI-low). That result splits the confluence family by ASSET PROFILE: it FAILS on high-beta/fat-tailed names (SOL 0.79, AVAX 0.34, XRP/DOGE/LTC no-edge) and CLEARS on smooth, MODERATE-volatility large-cap L1s (BTC survivor 1.99; ADA 1.75). ATOM (Cosmos) is squarely in the ADA class: a liquid large-cap L1 with ~5y Binance USD-M history, moderate ~4-5% daily vol (not memecoin fat tails), and clean multi-week directional legs (the 2021 run to ~$44, the 2022–2024 secular decline, subsequent swings) that the confluence mechanism harvests. It is distinct from the queued DOT port (different L1 ecosystem, different trend timing) and applies the exact mechanism + vol-scaling that produced the ADA significance-clearing result. NOT a mean-reversion fade (L53), NOT a squeeze breakout (L54), NOT a channel breakout (outlier class), NOT a cross-sectional basket (L52), NOT calendar/session (both falsified net-losing this session), NOT a NON-PRICE-feed gate (L46), NOT options/COIN-M (L50/L51). Fills the long-short gap (13.5% vs 86% long-only). To match the ADA construction and resist overfit: single 4H bar type, 3 params, fixed fast-EMA, volatility-scaled sizing (part of the ADA winner). Risk profile: ~1.5% equity risk/trade via ATR stop; vol-scaled notional = risk_budget / (atr_mult * ATR_4H) capped at 25% of equity*leverage; 2x leverage cap (reads self.config.leverage). Exactly 3 tunable parameters (slow/fast EMA ratio, daily-trend EMA length, ATR trailing multiple).

Implements the hypothesis exactly -- the ADA-winning construction (single 4H bar type, fixed fast EMA, 3 tunables, volatility-scaled risk-anchored sizing, ATR trail plus trend-break exit, long-short) ported to ATOM on 14229 4H bars (2020-02-07 to 2026-08-05, ~6.5 years). No extra feeds to mis-wire; clamps mirrored verbatim in _param_bounds; leverage 2.0 consumed in the notional cap, which is a fraction of CURRENT equity so losses cannot compound (L15). MEASURABILITY IS EXCELLENT, which matters after the last two rejections: 473 round trips (~73/yr), so walk-forward windows are dense and a 15-day holdout expects ~3 trades; the real 365-day sandbox places 78 trades from 39 entries with no drops, and all six Layer-2 scenarios plus static analysis pass locally. RESULTS (replaying THIS EXACT CLASS, 0.10% round-trip fees): avg +0.47% per trade net of fees (3x the 0.15% viability floor), profit factor 1.19, win rate 28%, trade-return kurtosis 17.2 with the top-3 trades only 13.6% of gross profit (broadly distributed, no one-day mirage); yearly avg/trade 2020 -0.14%, 2021 +0.98%, 2022 +0.69% (bear), 2023 +0.71%, 2024 +0.63%, 2025 -0.58%, 2026 YTD +1.59% -- positive in 5 of 7 years. The RECENT window is the strongest part, which is unusual for this family and worth the reviewer's attention: trailing-365-day replay is +0.81%/trade and the sandbox year returns +15.9% with Sharpe 0.78, PF 1.94, max drawdown 8.3% and avg +2.21%/trade. Every setting inside the declared bounds is positive (avg +0.13% to +0.56%, PF 1.05-1.23); the defaults sit on the plateau's better half (trend_ema 180 = 30 days, with 240/300 giving +0.42%/+0.41% and 120 dropping to +0.13%), and I chose ema_ratio 3.0 over 2.0 despite the latter's higher average because 2.0 carries kurtosis 52 versus 17. THE HONEST CAVEAT ON THE HYPOTHESIS'S PREMISE: ATOM does NOT reproduce the ADA result. ADA's confluence cleared significance with an optimized Sharpe 1.75 and full-sample PF around 2.6; ATOM's same-construction PF is 1.19 with a 28% win rate, i.e. the edge is real and measurable but roughly a third the strength, so the 'ADA-class moderate-volatility L1' thesis is only weakly supported here -- ATOM's 2021 melt-up and 2022-2024 decline produce fewer clean multi-week legs than ADA's. Whether a PF-1.19 sleeve with a strong recent year is worth the optimization budget is the Analyst's call; the measurement itself is solid, which is more than the last three candidates could offer.

The 'ATOM is in the clean-trending ADA class' premise is falsified even harder than the DOT sibling — ATOM is a thin, choppy laggard with no edge and catastrophic liquidity. profit_factor is 0.9946 (< 1.0, gross-losing) with expectancy -$2.4/trade, and the headline +13% total_return is a mark-to-market mirage: end_unrealized_pct 14.17 means the entire gain is an un-exited open position at the last bar, so realized performance is flat-to-negative. Sharpe is 0.097 with a bootstrap CI-low of -0.567 (essentially zero, straddles zero deeply), far below the family's ~1.5 DSR-clearing bar. Most decisively, it is uninvestable on capacity: capacity_usd is only $77K with impact_cost_pct 113.7% — modeled market impact EXCEEDS gross PnL, so ATOM's Binance USD-M liquidity is too thin to trade this at any deployable size. max_drawdown 19.4% (CI to 68%), with a regime-lumpy record (2020 -13.8%, 2021 -2.3%, 2025 -10.0%) carried only by 2023/2026. This is worse than the DOT port on every axis (PF below 1.0, $77K capacity). Abandon at BACKTEST_REVIEW rather than spend the optimization budget.

Long-short, single-instrument, single-bar-type momentum-confluence trend follower on ATOMUSDT.BINANCE USD-M perpetual, 4H bars. All reads come from the same 4H series: a fixed 12-bar fast EMA, a slow EMA at 12*ema_ratio, and a daily-horizon trend EMA of 180 4H bars (= 30 days). calculate_signal returns (EMA_fast - EMA_slow)/ATR14 every bar -- a continuous, volatility-scaled 4H momentum read. ENTRY only when the sign of that read agrees with the daily-horizon trend (close above/below the trend EMA); disagreement means FLAT. EXIT on a 3x-ATR trailing stop from the best close since entry or on the close crossing the trend EMA against the position. Sizing is volatility-scaled and risk-anchored: qty = equity*1.5%/(3*ATR), hard-capped at 25% of equity x 2x leverage. Three tunables: ema_ratio, trend_ema, trail_atr_mult.

Analyst/backtest_review: judge whether the strategy clears the deflated-Sharpe gate at PF 1.19; the mechanism is validated only on ADA/BTC, and ATOM lands closer to the SOL/DOT modest tier. Weigh this against the unusually favourable recent window (below).

Modest full-sample edge — the 'ADA-class L1' thesis is only weakly supported, an analyst/backtest_review call rather than a code defect. The developer honestly reports ATOM does NOT reproduce the ADA significance-clearing result: full-sample PF is 1.19 with a 28% win rate and avg +0.47%/trade, versus ADA's optimized Sharpe 1.75 / PF ~2.6. The bootstrap Sharpe CI [-0.89, 2.24] still straddles zero (PSR 0.83), and full-sample kurtosis is 17.2 (23.8 in the sandbox), though top-3 concentration is low (13.6%). Whether a PF-1.19 sleeve clears the deflated-Sharpe hurdle that rejected the 0.3-1.3 band repeatedly this session is the analyst's decision.

Backtest_review/analyst: this is the strongest-positioned confluence port on the recent window this session; give the strong trailing-year and OOS-regime placement due weight against the modest full-sample PF.

FAVOURABLE (noted for the analyst): unlike the DOT/SOL/AVAX siblings whose recent window was flat-to-negative, ATOM's RECENT window is its strongest — trailing-365-day replay +0.81%/trade, and the 365-day sandbox returns +15.9% with Sharpe 0.78, PF 1.94, max DD 8.3%, avg +2.20%/trade, positive in 5/7 years with 2026 YTD +1.59%. The walk-forward OOS windows and 15-day holdout therefore sit in a favourable rather than a faded regime, which is the opposite of the recent-regime risk that weighed against the other ports.

For live deployment, reconstruct _side/_extreme/_entry_atr from cache.positions_open() rather than the current signal.

should_exit() infers _side from the live signal sign on restart (_side==0) and re-seeds _entry_atr/_extreme from current values. Unreachable in backtest (state persists); only a live mid-position crash-restart risk, where it could pick the wrong side or reset the trailing reference.

469 trades over ~5.5y is a measurable sample; decorrelated (beta 0.010)

profit_factor 0.9946 < 1.0 (gross-losing), expectancy -$2.4/trade; the headline +13% total_return is 14.17% open-position unrealized MTM — realized is flat-to-negative

Sharpe 0.097 with CI-low -0.567 — essentially zero, straddles zero deeply; far below the family DSR bar

Catastrophic capacity: capacity_usd $77K with impact_cost_pct 113.7% — modeled impact EXCEEDS gross PnL; untradeable at any scale

max_drawdown 19.4% (CI to 68%); regime-lumpy (2020 -13.8%, 2021 -2.3%, 2025 -10.0%), carried only by 2023/2026

The 'ATOM is in the ADA class' premise is falsified — it is a thin, choppy laggard

AtomBinanceDualTimeframeMomentumConfluenceLS applied the ADA/BTC-validated confluence mechanism to ATOM on the thesis that it was a moderate-vol large-cap L1 in the same clean-trending class. The fair test refuted that on every axis: profit factor 0.9946 (gross-losing), Sharpe 0.097 with a CI deep into negative territory, a +13% headline that was purely an un-exited open position, and — decisively — $77K capacity with 113.7% impact cost, meaning modeled market impact exceeds gross PnL. The analyst abandoned it at backtest review as worse than the DOT port and uninvestable at any scale, a thin choppy laggard rather than an ADA-class name. It never reached optimization, analysis, or risk review.

Asset-class grouping by narrative must be validated, and capacity is a first-order gate — ATOM's profit factor below 1.0, a headline return that was entirely unrealized open-position MTM, and a modeled market impact exceeding gross PnL ($77K capacity) make it untradeable regardless of any signal, placing it firmly with the choppy laggards rather than the ADA class.

The analyst abandoned it at backtest review: the premise that ATOM belongs to the clean-trending ADA class was falsified even harder than the DOT sibling — ATOM is a thin, choppy laggard with a sub-1.0 profit factor, a Sharpe straddling zero far below the family's ~1.5 DSR bar, and a realized record flat-to-negative once the open-position MTM is excluded. Most decisively it is uninvestable: $77K capacity with impact exceeding gross PnL means the strategy cannot be traded at any deployable size.

A long-short, single-instrument dual-timeframe momentum-confluence trend follower on ATOMUSDT.BINANCE USD-M 4H (3 parameters) that aligned a fast/slow EMA spread with a 30-day trend EMA — all from one bar series — going flat on disagreement and trailing winners with an ATR stop, betting ATOM belonged to the clean-trending 'ADA class' where the confluence mechanism had cleared significance.

The backtest (ATOMUSDT.BINANCE 4H, 2371 data days from 2020-02) was essentially edgeless and untradeable: profit factor 0.9946 (gross-losing), expectancy -$2.4/trade, Sharpe 0.097 (CI-low -0.567), max drawdown 19.4% (CI to 68%) over 469 trades. The headline +13% total return was a mark-to-market mirage (end_unrealized_pct 14.17 — the entire gain an un-exited open position), and capacity was catastrophic at $77K with impact cost 113.7% (modeled impact exceeds gross PnL).