BtcSpotDailyDowHigherHighTrendLong

BTC BINANCE_SPOT Daily Dow-Theory Higher-High Trend Confirmation Long-Only (4-Parameter Structural Pattern, OHLCV-Only)

Long-only structural-pattern trend-confirmation strategy on BTCUSDT.BINANCE_SPOT using daily bars. This mechanism class is structurally distinct from every existing portfolio entry: it operates on the Dow Theory (Charles Dow, ~1900) definition of an uptrend — a SEQUENCE of higher highs — rather than on numerical breakouts (LINK Spot, DOGE 30-bar), pullbacks (ETH Spot), trend-strength momentum (TSM stack), volatility compression (BNB Squeeze), or session-range expansion (HL BTC ORB). The mechanism: track the 7-day rolling high, and require today's 7-day high to be GREATER than the 7-day high observed 7 days ago. This is a 'higher-high' confirmation pattern — the structural definition of an uptrend in classical Dow Theory. A SMA(50) trend filter provides the long-term bias gate (avoid bear markets). The mechanism is fundamentally about PATTERN RECOGNITION (higher peaks → uptrend), not numerical thresholds. Why this is a meaningful addition to the factory: existing strategies overwhelmingly use threshold-based signals (breakout above N-bar high, return > X sigma, BB compression < 25th percentile, etc). Threshold-based signals are vulnerable to regime decay because participants can adapt around them. PATTERN-BASED signals (higher highs, higher lows) capture structural market behavior that's more robust to participant adaptation — the pattern is the trend, not a side effect of one. BTC SPOT is chosen because: (a) BINANCE_SPOT remains at 10.2% vs 15% quota target, (b) BTC has the deepest spot liquidity on Binance (~$2B+ daily) eliminating ALL liquidity-floor concerns that killed ETC/DOGE 4H, (c) BTC spot has continuous Binance data since 2017 (8.5+ years) — the longest training history available for any crypto asset, allowing robust 3-window walk-forward, (d) BTC has the cleanest historical higher-high patterns due to its market-cap dominance and reduced sensitivity to alt-narrative cycles, (e) spot venue forces long-only which matches BTC's structural positive drift. Pure OHLCV — NO supplementary data dependency. 4 parameters total: h_lookback=7, sma_filter=50, atr_stop_mult=2.5, hard_stop_atr=1.8. NOT a cross-venue port (the HL SOL TSM abandon proved cross-venue trend-following claims are unreliable). NOT a cross-instrument port from a Binance-perp success. This is a freshly-motivated single-asset single-venue test of a classical mechanism not yet represented in the factory.

Implements the classical Dow Theory uptrend definition (a sequence of higher highs) as a pattern-recognition mechanism structurally distinct from the threshold-based breakouts, pullbacks, TSM, BB-squeeze and ORB strategies already in the factory. Rather than a numerical 'break above N-bar high' threshold (which participants can adapt around), it confirms a higher PEAK relative to the prior peak — the structural signature of a trend — and gates it with an SMA(50) bull filter to avoid bear markets. The pattern is expressed as a continuous ATR-normalized margin so it carries conviction and gives the optimizer a tunable confirmation strength rather than a binary trigger. Venue is BINANCE_SPOT because the strategy is strictly long-only and unleveraged with no funding edge, so the CASH fee model (~0.20% round-trip) is correct and leverage stays a true 1.0; BTC matches spot's structural positive drift and has the deepest spot liquidity (no liquidity-floor risk) plus 8.5+ years of continuous data for robust 3-window walk-forward. The dual-stop exit (hard entry stop + ratcheting chandelier) caps initial loss while letting confirmed uptrends run. This is a freshly motivated single-asset single-venue test, NOT a cross-venue/instrument port. Verified locally: Layer 1 static (clean, all divisions guarded by ATR>0/price>0/sma>0 checks), Layer 2 synthetic (all 6 scenarios pass with a non-frozen varying signal), and an uptrend simulation confirming the higher-high signal fires positively (99/120 bars) above the SMA gate so Layer 3 will produce trades.

Failed deflated Sharpe: DSR=0.0676, expected-max=6.80 over 225 trials — the optimized Sharpe of 4.63 is BELOW the best-of-N luck bar and statistically indistinguishable from selection noise (is_significant=false, PBO 0.6455 > 0.5). Walk-forward is_overfitted=true with avg IS 7.12 collapsing to avg OOS -0.569 and OOS windows [-1.43, -0.28, 0.0] — not a single positive out-of-sample window. The holdout 'passed=true' is a DEGENERATE artifact: a positive holdout Sharpe (2.08) measured against a NEGATIVE WF-OOS reference (-0.569) with ratio 0 trips a meaningless pass and must not override the failed DSR/PBO/OOS gates. The favorable full-period sharpe_ci_low (1.07) and PSR (0.9993) do not correct for the 225-trial selection that DSR does. This confirms the concerns flagged at pre-optimization: the headline rests on per-day-attribution-distorted metrics (annualized_volatility 615% makes the Sharpe an inflated artifact), a thin 45-trade sample, a 61.7% max drawdown, and bull-outlier concentration (optimized 2021 +465%, 2024 +147%, negative 2022/2025) — so the 'pattern-based, regime-resistant' thesis is not borne out; every genuine out-of-sample window is negative. Not a tunable-region case: the 225-trial sweep already produced negative average OOS across all windows and the only sensitivity cliff is the trivial spot leverage cap (position_fraction>1.0), so a second pass hits the identical wall with no robust region to tune toward. Not revise_hypothesis: there is no proven Dow-higher-high sibling to redirect to and the failure is overfit/bull-outlier dependence rather than structural decay of a working mechanism — reframing onto another instrument would just re-overfit. FAILURE PATTERN: single-asset long-only daily Dow higher-high trend on BTC spot produces a strong in-sample optimized Sharpe carried by 2021/2024 bull outliers and per-day-attribution-inflated volatility, but fails DSR (0.07)/PBO with all-negative walk-forward OOS and only a degenerate (negative-reference) holdout pass.

Long-only Dow-Theory higher-high trend-confirmation strategy on BTCUSDT.BINANCE_SPOT daily bars. Each bar compares the current h_lookback (7)-day rolling high against the rolling high from h_lookback days earlier: when today's peak exceeds the prior peak, a higher-high uptrend pattern is confirmed. The signal is the higher-high margin measured in ATR units = (current 7-day high - past 7-day high)/ATR (continuous, positive when a higher peak forms). It goes LONG when (a) the higher-high pattern is confirmed (signal > min_hh_atr) and (b) close > SMA(sma_filter=50) as a bull-regime bias gate. Exits use two ATR stops: a hard stop at entry - hard_stop_atr (1.8) x ATR for catastrophic protection, and a chandelier trailing stop at highest_close_since_entry - atr_stop_mult (2.5) x ATR to lock in trend gains. Long-only (SPOT CASH cannot short), pure OHLCV, leverage 1.0, ~95% capital deployed per position.

Functions and matches hypothesis: higher-high pattern + SMA(50) gate, long-only spot, ATR chandelier + hard stop; clean execution (0 dropped), metrics_reliable=true

Fresh classical mechanism (Dow higher-high) not yet in the factory; BTC spot is maximally liquid (capacity $347M) with 8.5yr history

Profit factor 1.61, omega 2.35 — positive expectancy with right-skewed trend-capture profile

Thin sample: 63 trades over 8.5 years (~7.4/yr) — borderline for 3-window walk-forward (~10 OOS trades/window)

Per-day-attribution-distorted metrics: annualized_volatility 473%, downside_deviation 63 — headline Sharpe 3.52 / Sortino 26.6 are inflated artifacts, not honest risk-adjusted figures

max_drawdown 71.4% (CI high 82%) — likely exceeds any promotion DD threshold; the SMA gate does not protect much in bears

Bull-outlier-concentrated (2017 +125%, 2021 +265%, 2024 +149%) with negative 2018/2022/2025 and a weak recent regime (2025 -14%, 2026 -0.4%) where the holdout lands

Clean execution (45-63 trades, 0 dropped); fresh classical mechanism; BTC spot liquid

In-sample optimized headline strong (PF 1.99) — but not validated out-of-sample

Failed deflated Sharpe: DSR=0.0676, optimized Sharpe 4.63 BELOW the expected-max luck bar 6.80 over 225 trials (is_significant=false)

PBO=0.6455 (>0.5) — selection more likely than not overfit

Walk-forward is_overfitted=true: avg IS 7.12 collapses to avg OOS -0.569, windows [-1.43, -0.28, 0.0] — none positive

Holdout 'passed=true' is degenerate: positive holdout (2.08) over a NEGATIVE WF-OOS reference (-0.569), ratio 0 — not a real validation

Per-day-attribution-distorted metrics (annualized_volatility 615%) — headline Sharpe is an inflated artifact; max drawdown 61.7%

Bull-outlier concentrated (optimized 2021 +465%, 2024 +147%); thin 45-trade sample

BtcSpotDailyDowHigherHighTrendLong introduced a fresh classical mechanism — Dow Theory higher-high confirmation gated by a 50-SMA — on maximally-liquid BTC spot, argued to be more robust to participant adaptation than threshold breakouts. It cleared the pre-optimization gate on an attractive headline (PF 1.61, +1164%), but reviewers flagged the thin 63-trade sample, a 71% drawdown, per-day-attribution-inflated Sharpe, and bull-outlier concentration. Optimization confirmed the concern: deflated Sharpe 0.07, PBO 0.65, an in-sample 7.12 collapsing to an all-negative OOS average, and a meaningless degenerate holdout — the regime-resistance thesis unsupported. The analyst judged it overfit/bull-outlier-dependent with no robust tuning region and no proven sibling, and abandoned it; it reached optimization and analysis but was abandoned after a single iteration, never advancing to risk review.

A 'pattern-based, regime-resistant' Dow higher-high trend on BTC spot is not borne out: the strong in-sample optimized Sharpe is carried by 2021/2024 bull outliers and per-day-attribution-inflated volatility, but every genuine out-of-sample window is negative — favorable full-period CI/PSR do not correct for 225-trial selection, and a degenerate holdout pass over a negative reference must not override failed DSR/PBO/OOS gates.

After passing the backtest-review gate (verdict: optimize), the analyst abandoned it at the post-optimization stage: it failed deflated Sharpe (DSR 0.07 with optimized Sharpe 4.63 below the expected-max luck bar 6.80 over 225 trials, is_significant=false, PBO 0.65), the walk-forward was overfit (avg IS 7.12 collapsing to avg OOS -0.569 with no positive window), and the holdout 'pass' was a degenerate positive-over-negative-reference artifact.

A long-only Dow-Theory higher-high trend-confirmation strategy on BTCUSDT.BINANCE_SPOT daily bars (OHLCV only) that confirmed an uptrend when the current 7-day rolling high exceeded the 7-day high from 7 days earlier (a higher peak) while price was above its 50-day SMA, exiting on a 2.5× ATR chandelier trailing stop or a 1.8× ATR hard stop — a pattern-recognition mechanism pitched as more regime-robust than threshold breakouts.

The base backtest over 2017-08-17 to 2026-05-18 showed strong-looking but distorted results across 63 trades — total return +1164%, CAGR 34.5%, profit factor 1.61, Sharpe 3.52 (per-day-attribution inflated, annualized vol 473%), a 71.4% max drawdown, 37% win rate — concentrated in 2017/2021/2024 bull years; optimization (Sharpe 4.63, +3281%) kept a healthy in-sample headline that collapsed out-of-sample.