DogeSpotMultiDayDrawdownAccumulationLong

DOGE Spot Multi-Day Drawdown Accumulation Long (BINANCE_SPOT, Daily, OHLCV-Only)

A long-only, single-instrument, single-venue, OHLCV-ONLY strategy on DOGEUSDT.BINANCE_SPOT DAILY bars. Same proven drawdown-accumulation mechanism as the landed BtcSpotDrawdownAccumulationLong + AvaxSpotDrawdownAccumulationLong + XrpSpotDrawdownAccumulationLong: buy when DOGE has dropped meaningfully from its 30-day high AND is still above its 100-day SMA; sell when DOGE recovers to within 5% of the peak, hits a stop, or times out. DOGE fills a unique cycle profile: social-sentiment / memecoin-narrative driven (Elon tweets 2021, Twitter rebranding 2023, election cycles) with distinct drawdown-bounce timing vs BTC/AVAX/XRP. Higher-volatility (~90% annualized) means larger drawdown thresholds and bigger recoveries. DOGEUSDT.BINANCE_SPOT has been listed since 2019 — ~6 years of daily bars. Five parameters total. Hits BINANCE_SPOT venue quota gap (currently 6.9% vs ≥15% target).

Replicates the landed BTC/AVAX/XRP spot drawdown-accumulation mechanism on DOGE, which has a distinct memecoin/social-sentiment drawdown-bounce cycle and higher (~90%) volatility, justifying a larger 30% drawdown threshold and bigger recoveries. calculate_signal returns a continuous, bar-varying drawdown depth in [0,1) rather than gating to 0.0, so the verifier sees real signal variation; the 100-SMA-above regime is stored as self._regime_ok and ANDed in should_enter so accumulation only happens while the structural uptrend is intact (avoids catching a falling knife in a bear market). Exits mirror the entry edge: a recovery exit when the drawdown closes to within 5% of the peak (the mean reversion of the drawdown has played out), a 30% hard stop for the terminal-decay tail the SMA filter is meant to exclude, and a 60-bar time stop. Venue routing is correct for the strategy: long-only, no leverage, and the edge is not funding-based, so BINANCE_SPOT (CASH, leverage hard-capped at 1) is the right venue and fills the spot-venue quota gap; should_enter only ever returns BUY (spot cannot short) and sizing is plain equity*fraction/price with no leverage. DOGEUSDT.BINANCE_SPOT has ~6 years of daily history. Targets (recoveries toward the peak) are far larger than the ~0.20% spot round-trip fee, keeping the strategy fee-viable.

Failed deflated Sharpe: DSR=0.0438 (vs 0.95 bar), expected-max=13.65 over 225 trials — the selected optimized Sharpe of 6.96 is FAR BELOW the best-of-N luck bar, so the eye-catching headline is statistically below the noise ceiling (is_significant=false). Every other generalization gate also fails: PBO 0.7249 (72% overfit probability), sharpe_ci_low -3.1881 (CI straddles 0 from -3.19 to +14.67), walk-forward is_overfitted=TRUE with avg IS 5.40 collapsing to avg OOS 0.0 and ALL three OOS windows producing ZERO trades, a FAILED holdout (holdout_sharpe 0.0, zero trades), AND a FAILED sensitivity with 2 cliffs on core params (position_pct: 4.46->0.0 at 1.045/1.14; hard_stop_pct: 4.46->0.15 at 24 — the heatmap shows whole dead rows). Root cause is twofold: (1) structural sparsity — 15 base / 10 optimized trades over ~6 years leaves the walk-forward OOS and holdout windows empty (zero-trade collapse); and (2) outlier/mania concentration — the headline Sharpe is carried by ~3 DOGE memecoin-pump days (2021-01-28 +257.5%, optimized 2021-04-13 +119%, +64% Apr 2021) with returns concentrated entirely in 2021 (base +200%, optimized +642%) and degenerate dispersion (annualized_volatility 1201%, downside_deviation 172). Not iterate (attempt 1 of 2): with zero trades across all three OOS windows AND the holdout, no parameter change manufactures out-of-sample occurrences from a structurally rare, mania-driven trigger; the 15-trade base is already the densest config, and the two core cliffs (sizing + stop) confirm no robust region exists, so a second best-of-225 sweep re-finds another 2021-pump-fit. Not revise_hypothesis: the drawdown-accumulation mechanism is already captured by the landed BTC/AVAX/XRP spot siblings — DOGE is the outlier/mania-driven arm of that family's asset-selectivity, not a proven mechanism stranded on a once-working, now-dead target. FAILURE PATTERN: transplanting the proven spot drawdown-accumulation mechanism onto a high-volatility memecoin (DOGE) produces a flashy raw Sharpe (4.46 base / 6.96 optimized) that is purely an artifact of a handful of 2021 mania pump days — after best-of-225 selection it fails deflated Sharpe (0.044, below a 13.65 luck bar), PBO 0.72, is_overfitted with zero-trade OOS, a zero-trade holdout, and sensitivity cliffs on sizing/stop. A sparse drawdown-bounce signal on a mania-concentrated meme asset is structurally unvalidatable out-of-sample, the same sparsity + outlier-luck signature as the abandoned ETH Bollinger and BNB trend-continuation siblings; the PSR=0.96 / DSR=0.04 split and the 1201% annualized vol are the diagnostic tells.

Long-only multi-day drawdown accumulation on DOGEUSDT.BINANCE_SPOT DAILY bars, OHLCV only. Buys when DOGE is >=30% below its rolling 30-day high while still above its 100-day SMA (structural uptrend intact); sells when DOGE recovers to within 5% of the peak (drawdown closed) while in profit, hits a 30% hard stop, or times out after 60 bars. The signal is the continuous fractional drawdown from the 30-day high (recomputed every bar); the 100-SMA regime is a separate boolean entry gate. Spot CASH account, long-only (never sells to enter), leverage 1, sized at 95% of equity notional.

Proven mechanism: identical drawdown-accumulation logic to the landed BTC/AVAX/XRP spot siblings; code is correct and faithful to the hypothesis (15 longs, no shorts, spot CASH semantics)

Returns are real DOGE moves, not a sizing artifact: avg_position_pct 95.5% matches config, end_unrealized_pct 0.0, and the +257% day is the genuine Jan-2021 DOGE pump

Recent years both strongly positive (2024 +64%, 2025 +58%) — edge is not purely a 2021 relic

Sparse trade count (15) is in-family for the rare >=30%-drawdown trigger that already validated on siblings

max_drawdown 84.4% at 95% position sizing — catastrophic risk that threatens eventual promotion

annualized_volatility 1201% and Sharpe 4.46 with CI [-5.09, 10.69] — Sharpe is uninterpretable; result dominated by a handful of extreme days

15 trades splits to ~5 per walk-forward window — thin sample, and the 2021 annual (+200%) is essentially one outlier day, so DSR/holdout may reject it

Negative in 2020, 2022, 2023 — outside the mania and 2024/2025, the edge is mixed

Mechanism is proven on siblings (landed BTC/AVAX/XRP spot drawdown-accumulation longs); code correctly implements the continuous-drawdown signal with a separate SMA regime gate (avoids the gated-to-0.0 anti-pattern) and respects spot long-only/CASH semantics.

DOGEUSDT.BINANCE_SPOT data spans ~6 years and the strategy trades as specified (15 base / 10 optimized long entries, recovery/stop/time exits).

Deflated Sharpe = 0.0438 (is_significant FALSE): optimized Sharpe 6.96 is FAR below the expected-max luck bar of 13.65 over 225 trials — below the best-of-N noise ceiling.

PBO = 0.7249 (~72% probability of backtest overfitting, > 0.5).

sharpe_ci_low = -3.1881 (ci_high 14.6713) — an enormous bootstrap CI straddling 0; the Sharpe point estimate is not distinguishable from noise.

Walk-forward is_overfitted = TRUE: avg IS 5.40 collapses to avg OOS 0.0, with ALL three OOS windows = [0.0, 0.0, 0.0] (zero trades out-of-sample).

Holdout FAILED: holdout_sharpe 0.0, ratio 0 (zero trades in the holdout window).

Sensitivity FAILED with 2 cliffs on CORE params: position_pct (4.46 -> 0.0 at 1.045/1.14) and hard_stop_pct (4.46 -> 0.15 at 24); heatmap shows whole dead rows — a knife's edge, not a plateau.

Degenerate / outlier-driven metrics: annualized_volatility 1201%, downside_deviation 172, headline carried by ~3 memecoin-mania days (2021-01-28 +257.5%, optimized 2021-04-13 +119%); returns concentrated entirely in 2021 (base +200%, optimized +642%) with other years negative/absent.

DogeSpotMultiDayDrawdownAccumulationLong copied the drawdown-accumulation logic of the landed BTC, AVAX, and XRP spot strategies onto DOGE, betting that its higher volatility and memecoin cycle would yield bigger recoveries. The backtest looked spectacular on the surface — a 4.46 Sharpe and +147% return rising to a 6.96 Sharpe after optimization — and it passed backtest review into full optimization. But those numbers rested on ~3 DOGE pump days in 2021 amid 1201% annualized volatility, and every forward gate collapsed: deflated Sharpe 0.044 against a 13.65 luck bar, 72% overfitting probability, zero trades in all out-of-sample and holdout windows, and sensitivity cliffs on its core sizing and stop parameters. The analyst abandoned it as the mania-concentrated, structurally-sparse arm of an asset family whose edge is already captured by its validated siblings — the same outlier-luck signature seen in the abandoned ETH Bollinger and BNB trend-continuation strategies.

Transplanting a proven mechanism onto a high-volatility, mania-driven meme asset yields a flashy raw Sharpe that is pure outlier luck; a sparse drawdown-bounce trigger (15→10 trades over 6 years) cannot populate walk-forward or holdout windows and so is structurally unvalidatable out-of-sample.

It cleared the backtest-review gate (verdict: optimize) but the analyst abandoned it after optimization for failing every generalization gate: deflated Sharpe 0.044 vs a 13.65 expected-max luck bar over 225 trials, PBO 0.72, sharpe_ci_low -3.19, walk-forward is_overfitted=true with all three OOS windows and the holdout producing zero trades, plus a failed sensitivity test (2 cliffs on position_pct and hard_stop_pct).

A long-only, spot, OHLCV-only DOGEUSDT daily drawdown-accumulation strategy — the same mechanism as the landed BTC/AVAX/XRP spot siblings — buying when DOGE fell ≥30% from its 30-day high while still above its 100-day SMA, then exiting on recovery to within 5% of the peak, a hard stop, or a time stop.

The base backtest took 15 long trades over ~6 years with a 53.3% win rate, profit factor 1.28, a +146.99% total return, and a headline Sharpe of 4.46 — but with 1201% annualized volatility, an 84.4% max drawdown, and a Sharpe CI of [-5.09, 10.69]. Optimization thinned it to 10 trades with a higher Sharpe (6.96), profit factor 2.72, and lower max drawdown (52.3%), yet the gains stayed concentrated in a handful of 2021 mania days.