BnbTrendQualityGatedTimeSeriesMomentumLS1D

BNB USD-M Daily Trend-Quality-Gated Time-Series Momentum, Long-Short (Single-Instrument BINANCE Perp, Daily Bars, Validated-Family Edge Applied to a Decorrelated, Trend-Persistent Major, Flat-in-Chop ADX Gate, Low-Parameter)

A LONG-SHORT, SINGLE-INSTRUMENT, DAILY time-series (absolute) momentum strategy on BNBUSDT.BINANCE (USD-M perpetual) that deliberately applies the factory's ONE repeatedly-validated, promoted edge — daily trend-following on cleanly-trending majors (BTC/ETH/SOL TS-momentum siblings) — to a major that the portfolio does NOT yet cover with a long-short trend book and that is structurally DECORRELATED from those three. BNB is an exchange-token / BNB-Chain asset whose return drivers (exchange volume, ecosystem, buyback/burn) differ from the pure L1 betas of BTC/ETH/SOL, so a trend book on BNB adds genuine diversification rather than the correlated family risk the analyst flagged on cross-sectional re-expressions of the same majors universe. It is built to PASS where the LTC trend extension FAILED: the LTC meta-note's lesson was to screen for actual trend persistence and to not let a naked trend-cross get chopped on a range-bound asset. BNB has demonstrably produced sustained directional moves (e.g. the 2021 and 2023-2024 advances) — unlike laggard/range-bound LTC — AND this design adds a TREND-QUALITY GATE (ADX/efficiency filter) that keeps the book FLAT during choppy regimes, the exact failure mode (16.8% win-rate whipsaw) that wiped LTC. It avoids every engine and edge landmine confirmed this session: SINGLE-INSTRUMENT / SINGLE-VENUE / SINGLE-ACCOUNT (no multi-instrument equity/return aggregation artifact that corrupted the pairs / rotation / sleeve / cash-carry runs); DAILY bars (not the intraday-breakout/impulse/vol-expansion variants that are chop-dominated and negative on majors); no options, no funding-carry, no cross-venue. BNB is a deep-liquidity major (long history for robust walk-forward/holdout) and is far less single-event-gap-fragile than regulatory alts like XRP. Low parameter count (≤4 active knobs) to resist overfitting.

Applies the factory's one repeatedly-validated, promoted edge — daily TS-momentum on cleanly-trending majors (the BTC/ETH/SOL TS-momentum siblings) — to BNB, a major the portfolio does not yet cover with a long-short trend book and that is structurally decorrelated (exchange-token / BNB-Chain return drivers vs pure-L1 betas), so it adds diversification rather than the correlated family risk the analyst flagged on cross-sectional re-expressions of the same universe. It is built to PASS where the LTC trend extension FAILED: the LTC lesson was that a naked trend-cross gets chopped on a range-bound asset (16.8% win-rate whipsaw). BNB has demonstrably produced sustained directional advances (2021, 2023-2024), and the ER trend-quality gate explicitly keeps the book flat in chop — neutralizing that exact failure mode. It avoids every engine/edge landmine from this session: SINGLE-INSTRUMENT / SINGLE-VENUE / SINGLE-ACCOUNT (no multi-instrument equity-aggregation artifact), DAILY bars (not chop-dominated intraday breakout/impulse/vol-expansion variants), no options, no funding-carry, no cross-venue. Long+short on a USD-M perp => BINANCE futures venue (shorts required); leverage=1.0 with per-side notional <= 40% of equity, so sizing never references leverage (no unused-leverage gate). Daily round-trip taker cost ~0.10% is dwarfed by multi-week trend captures, and turnover is low. Low parameter count (4 active knobs: lookback, er_threshold, entry_threshold, stop_loss_pct) resists overfitting for a robust walk-forward/holdout on BNB's long, deep-liquidity history.

Outlier-driven mirage on a suspect bar, too sparse to optimize. The +210% total return is entirely one day (2021-03-31 +218%, largest_win $198.7k); strip it and 2020/2022/2023/2026 are negative and 2024/2025 only modestly positive (skew 26, kurtosis 698 confirm single-outlier dominance). That +218% single-day return is not explainable by the strategy's own 0.4x notional sizing — BNB's largest real daily moves are ~30-40%, so a book at ~50% avg position cannot gain 218% in a day; combined with a backtest start (2020-02) that predates the BNBUSDT USD-M perp listing (~Feb 2021), this is a suspected data/sizing artifact rather than captured edge (symptom stated; mechanism not independently verified — recommend checking the early BNB perp data and a per-trade notional cap before any retry). Independently, the strategy is statistically too thin to optimize: 25 trades over 6 years cannot support a 4-parameter, 3-window, 225-trial walk-forward (the optimizer will curve-fit to the single 2021 day), and it shows negative alpha (-0.014) and information_ratio (-0.75) with a Sharpe CI that straddles 0 — it's mostly long-BNB beta, not a differentiated trend edge. Optimization tunes parameters around an edge; it cannot manufacture one from a single implausible bar. FAILURE PATTERN: single-instrument daily trend-following on a major whose entire backtest return is one parabolic outlier day, with too few trades (~25) for walk-forward and negative risk-adjusted alpha vs buy-hold — a high headline return masking an outlier-dependent, beta-driven book, compounded by a suspect single-day magnitude that the strategy's own sizing cannot produce.

Single-instrument, single-venue daily LONG-SHORT time-series (absolute) momentum on BNBUSDT.BINANCE (USD-M perpetual). The signal is a Sharpe-like trend t-stat (trailing lookback-day return / realized daily-return vol), tanh-normalized to (-1,1). A Kaufman Efficiency-Ratio trend-quality gate (|net move| / total path over the lookback) keeps the book FLAT during choppy/range-bound regimes and only allows entries when the move is efficient (ER >= er_threshold). Enters long/short past +/-entry_threshold while flat, exits on direction flip beyond +/-exit_threshold (hysteresis) or a catastrophic stop. Daily bars, no take-profit (lets trend run), 4 active knobs.

Clean single-instrument/single-venue/single-account design that avoids the multi-leg accounting artifacts flagged elsewhere

The ER chop-gate concept is sound and holding period (~39d) is consistent with a genuine trend-follower

Low fee drag (commission 0.85% of gross) — not a fee-fragility case

Entire +210% headline is ONE day: 2021-03-31 = +218% (largest_win $198.7k). Every year except 2021 is flat-to-negative; skew 26 / kurtosis 698 confirm single-outlier dominance

That +218% single-day return is implausible for the stated 0.4x notional sizing (BNB's real daily moves are ~30-40%) — suspected data/sizing artifact, likely tied to the pre-Feb-2021 window before BNBUSDT USD-M perp existed (backtest starts 2020-02 but the perp listed ~2021)

Only 25 trades over ~6 years — too sparse for a meaningful 4-param / 225-trial walk-forward; the optimizer would fit to the single 2021 day and OOS windows would be near-empty

No differentiated edge: alpha -0.014, information_ratio -0.75, benchmark_correlation 0.71 — it's largely long-BNB beta and underperforms buy-hold risk-adjusted; Sharpe 0.46 with CI low -2.61 straddling 0

BnbTrendQualityGatedTimeSeriesMomentumLS1D applied the validated daily-trend edge to BNB as a decorrelated major, adding a Kaufman efficiency-ratio gate to keep the book flat in chop and avoid the whipsaw that wiped LTC. The design was clean and low-fee, but the entire +210% return came from a single 2021-03-31 +218% day — implausible for its 0.4x sizing and likely tied to a backtest window predating the BNB perp listing — leaving every other year flat-to-negative with negative alpha and mostly long-BNB beta. The analyst ruled it an outlier-driven mirage on a suspect bar, too thin at 25 trades to support walk-forward, recommending a data/sizing check before any retry. It ended after one iteration as abandoned, never advancing to optimization or risk review.

A high headline return that is entirely one parabolic day is not an edge: 25 trades over 6 years with skew 26 / kurtosis 698, negative alpha, and 0.71 beta correlation is an outlier-dependent, long-beta book, not a differentiated trend edge — and a +218% single-day return the strategy's own sizing cannot produce signals a likely pre-listing data artifact that must be checked (plus a per-trade notional cap) before any retry; the ER chop gate, though sound, made the signal too sparse to validate.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) as an outlier-driven mirage on a suspect bar — the +218% single day is implausible for the strategy's ~0.4x notional sizing (BNB daily moves are ~30-40%) and likely a data/sizing artifact from a backtest start (2020-02) predating the BNBUSDT USD-M perp listing (~Feb 2021) — and at just 25 trades it is far too sparse to optimize, so optimization and all later stages were never reached.

A single-instrument, single-venue long/short daily time-series momentum strategy on BNBUSDT.BINANCE USD-M perps — going long/short on a vol-scaled trailing-return trend score with hysteresis exits and a Kaufman Efficiency-Ratio chop gate that keeps the book flat in range-bound regimes — applying the factory's validated daily-trend edge to BNB as a decorrelated exchange-token major, explicitly adding the trend-quality gate to fix the chop-whipsaw that killed the LTC trend extension.

Over ~6 years and only 25 trades (16 long / 9 short, ~39-day holds, low fees at 0.85% of gross) the headline looked strong — total return +210%, profit factor 1.85, Sortino 5.08, Sharpe 0.46 (CI [-2.61, 1.03]) — but the entire result was one day (2021-03-31 +218%, largest win $198.7k), with skew 26 / kurtosis 698 confirming single-outlier dominance, every year but 2021 flat-to-negative, negative alpha -0.014, information ratio -0.75, and 0.71 benchmark correlation (mostly long-BNB beta).