DogeDailyStrongBodyTrendLong

DOGE Daily Strong Body Trend Continuation Long

A minimalist long-only single-instrument trend-continuation strategy on DOGEUSDT perpetual futures using daily (1D) bars. Uses ONLY OHLCV data — no funding, no L/S ratios, no liquidations, no pairs — to keep verification fast and parameter count minimal. The hypothesis: when a daily candle closes with a strong bullish body (body size as a fraction of total range exceeds a threshold) AND the close is above a short-term SMA, this represents directional conviction by aggressive buyers and tends to persist for several days. DOGE has not appeared in any recent failure list and is not in the active pipeline. Daily timeframe means low trade frequency (~20-40 trades/year), so per-trade fee drag (~0.10% RT on futures) is minimal relative to expected multi-percent moves over 3-5 day holds.

Iteration 2 fix for the Layer-2 synthetic failure: _body_fraction was a @staticmethod taking one argument, but the verification proxy invoked it bound so that both self and the bar were passed (TypeError: takes 1 positional argument but 2 were given). The minimal change converts it to a normal instance method (def _body_fraction(self, b)), which matches the self._body_fraction(bar) call site exactly; all signal/entry/exit/sizing logic that already passed earlier layers is unchanged. The strategy still returns the continuous signed body fraction from calculate_signal (varies every bar) and gates entry on the SMA trend filter in should_enter.

Failed deflated Sharpe: DSR=0.0504 (vs 0.95 bar), expected-max=6.0872 over 225 trials — the optimized Sharpe of 4.22 is BELOW the best-of-N luck bar, statistically indistinguishable from selection noise (is_significant=false). The decisive multiple-testing/walk-forward gates reject it: PBO 0.5238 (>0.5) and walk-forward is_overfitted=TRUE with avg IS 4.64 collapsing to avg OOS 1.44 and OOS windows [0.0, -0.68, 5.00] (one zero-trade, one negative, propped entirely by a single 5.0 window). The optimizer over-thinned the 113-trade base down to 24 trades by raising body_threshold 0.6->0.86 and risk_fraction to 1.2, fitting the strong-body subset to a handful of big DOGE up-moves (base return_kurtosis 56.7, end_unrealized 197.6). The passing gates do not rescue it: holdout PASSED (4.53), sensitivity PASSED (0 cliffs), and sharpe_ci_low is positive (1.65), but a DSR of 0.05 far below the 6.09 luck bar plus is_overfitted=TRUE plus PBO>0.5 are each disqualifying — a holdout pass on a config whose walk-forward OOS averages 1.44 with a negative window and whose Sharpe sits below the best-of-N luck bar is in-sample selection luck, not forward evidence. The PSR 0.9999 / DSR 0.05 split is the diagnostic tell of best-of-225 inflation. Not iterate (optimization attempt 1): sensitivity is already clean (0 cliffs) so the optimizer is on its robust plateau with no under-explored region to tune toward, and it has already over-thinned to 24 trades with an overfit, negative-OOS-window walk-forward — a second best-of-225 sweep re-finds another thin, sub-luck-bar config. Not revise_hypothesis: the strong-body trend-continuation premise has no proven sibling stranded on a dead target, and DOGE is itself the outlier/bull-concentrated arm. FAILURE PATTERN: a minimalist long-only daily strong-body trend-continuation on a high-volatility memecoin (DOGE) optimizes to a flashy headline (Sharpe 4.22, PF 4.04, Sortino 15.5, holdout 4.53, 0 sensitivity cliffs, positive CI lower bound) yet still fails deflated Sharpe (0.05, optimized Sharpe below a 6.09 luck bar) with PBO 0.52 and is_overfitted=TRUE because best-of-225 selection thins 113 trades to 24 fit to a few big up-moves — the same PSR-high/DSR-near-zero, over-thinned signature as the abandoned DOGE drawdown-accumulation and SOL Keltner siblings. A passing holdout and clean sensitivity grid do not rescue an edge that the deflated-Sharpe gate identifies as best-of-N noise.

Minimalist long-only trend-continuation strategy on DOGEUSDT.BINANCE (USD-M perp), DAILY bars, OHLCV-only. Enters long when a daily candle closes with a strong bullish body (body/range > 0.6) AND the close is above a 20-day SMA (directional buyer conviction that tends to persist). Exits on a trend break (close back below the SMA), a 6% stop-loss, a 12% take-profit, or a 7-bar time stop. ~20-40 trades/year. Positions sized at 90% of account equity.

Verification failed (Layer 2 — synthetic scenarios): Parameters used: ['sma_period', 'max_hold_bars', 'risk_fraction', 'stop_loss_pct', 'body_threshold', 'take_profit_pct'] Check that __init__ sets all attributes from self.parameters.get(). - steady_uptrend: TypeError: DogeDailyStrongBodyTrendLong._body_fraction() takes 1 positional argument but 2 were given (bar timestamp: 1735690860000) - steady_downtrend: TypeError: DogeDailyStrongBodyTrendLong._body_fraction() takes 1 positional argument but 2 were given (bar timestamp: 1735690860000) - flat_ranging: TypeError: DogeDailyStrongBodyTrendLong._body_fraction() takes 1 positional argument but 2 were given (bar timestamp: 1735690860000) - volatility_spike: TypeError: DogeDailyStrongBodyTrendLong._body_fraction() takes 1 positional argument but 2 were given (bar timestamp: 1735690860000) - zero_volume: TypeError: DogeDailyStrongBodyTrendLong._body_fraction() takes 1 positional argument but 2 were given (bar timestamp: 1735690860000) - price_gap: TypeError: DogeDailyStrongBodyTrendLong._body_fraction() takes 1 positional argument but 2 were given (bar timestamp: 1735690860000)

113 long trades over ~6 years (~19/yr) matches the hypothesized 20-40/yr frequency; signal fires cleanly (113/113 signals submitted, 0 dropped)

Trades implement the stated mechanism: long-only, ~4.25-day avg hold (vs hypothesized 3-5 day holds), trend-continuation entries

Genuine functioning edge: PF 1.24, Sharpe 1.15 (CI low +0.04), Sortino 3.16, PSR 0.98

Adequate sample size for meaningful parameter sensitivity analysis

Returns heavily bull-concentrated (2021 +458%, 2023/2024 strong) with negative recent years (2022 -36%, 2025 -3%, 2026 -22%) — decay signature that may fail holdout/DSR post-optimization

max_drawdown 59.9% with risk_fraction 0.9 (avg_position_pct ~100%) is excessive — sizing needs tuning

end_unrealized_pct 197.6 and return_kurtosis 56.7: headline depends on open-position MTM and fat tails — to be scrutinized after optimization

Holdout PASSED (ratio 3.143, holdout Sharpe 4.53)

Sensitivity PASSED with 0 cliffs — smooth parameter surface

sharpe_ci_low positive (1.6455) and optimized annual returns positive every year in-sample

Deflated Sharpe 0.0504 (is_significant FALSE); optimized Sharpe 4.22 is BELOW the expected-max luck bar of 6.09 over 225 trials — indistinguishable from best-of-N selection

Walk-forward is_overfitted=TRUE: avg IS 4.64 collapses to avg OOS 1.44, with OOS windows [0.0, -0.68, 5.00] (one zero, one negative, propped by one window)

PBO 0.5238 (>0.5) — selection more likely than not overfit

Optimizer over-thinned 113 trades to 24 by raising body_threshold to 0.86; edge fit to a handful of big DOGE up-moves (base kurtosis 56.7)

PSR 0.9999 vs DSR 0.05 split is the diagnostic tell of best-of-N inflation

DogeDailyStrongBodyTrendLong implemented a clean strong-body trend-continuation premise on DOGE daily bars, firing 113 long trades over ~6 years for a 1,258% return and earning an 'optimize' verdict at backtest review despite a bull-concentrated profile and a 59.9% drawdown. Optimization produced a dazzling headline — Sharpe 4.22, profit factor 4.04, a passing holdout (4.53), and zero sensitivity cliffs — but achieved it by raising the body threshold to 0.86 and thinning the sample to just 24 trades fit to a few large DOGE up-moves. The deflated-Sharpe gate flagged this as best-of-N noise (DSR 0.05 against a 6.09 luck bar), corroborated by an overfit walk-forward with a negative OOS window and PBO above 0.5. On its second iteration the analyst abandoned it, noting the same over-thinned, PSR-high/DSR-near-zero signature as earlier abandoned DOGE and SOL siblings.

A flashy optimized headline and a passing holdout do not survive the deflated-Sharpe gate when best-of-N selection over-thins the trade base (113→24) and fits the edge to a handful of big up-moves — the PSR-high (0.9999) / DSR-near-zero (0.05) split is the tell of selection noise.

Despite a passing holdout (ratio 3.143, Sharpe 4.53) and a clean sensitivity grid (0 cliffs), the analyst abandoned it post-optimization because the deflated Sharpe was 0.05 (vs the 0.95 bar) with the optimized Sharpe 4.22 sitting below the best-of-225 luck bar of 6.09, walk-forward was_overfitted=TRUE (avg IS 4.64 collapsing to avg OOS 1.44 with windows [0.0, -0.68, 5.00]), and PBO was 0.52.

A minimalist long-only, OHLCV-only daily trend-continuation strategy on DOGEUSDT perpetual futures — enter long when a daily candle closes with a strong bullish body (body/range above a threshold) while the close sits above a short-term SMA, exiting on a trend break, stop, take-profit, or time stop.

The initial backtest returned 1,258% total (CAGR 53%) over 113 trades with Sharpe 1.15 (CI low +0.04), profit factor 1.24, win rate 38.9%, but a 59.9% max drawdown. After optimization the headline looked far stronger (Sharpe 4.22, PF 4.04, Sortino 15.5, win rate 62.5%, max DD 18%) — but only because the optimizer thinned the trade count from 113 to 24.