ADA 4H Range-Outlier Bullish-Close Long (Single-Bar Range-Percentile Signal)

A long-only, single-instrument, single-timeframe strategy on ADAUSDT.BINANCE 4H bars (the most-reliably-fresh bar type per the persistent audit_stale pattern). Implements a NEW MECHANISM not yet tried in the portfolio: SINGLE-BAR RANGE-OUTLIER with BULLISH-CLOSE confirmation. The mechanism: detect bars where the bar's true range (high - low) is in the top quartile of its trailing 50-bar distribution AND the bar's close is in the upper third of the bar's own range. This identifies bars with unusually large range AND directional bias toward the high — combined signal that captures momentum-injection events (institutional accumulation, news-driven flow, liquidity absorption resulting in upward push). Critically distinct from every existing pipeline mechanism on ADA 4H: (a) NOT Donchian (uses single-bar range, not channel breach); (b) NOT ATR expansion (uses raw range percentile, not smoothed ATR ratio); (c) NOT multi-bar close-position consistency (single-bar, not averaged); (d) NOT bar-return z-score (uses RANGE size + close-within-range geometry, not return magnitude); (e) NOT channel-width expansion (single-bar range, not multi-bar channel); (f) NOT sequence-based (single-bar snapshot). The conjunction (unusual range + bullish close-position) provides natural noise filtering — a wide-range bar with the close near the LOW would NOT trigger (sellers won), while wide-range bar with close near HIGH does trigger (buyers won). Expected trade frequency: ~35-55 trades/year on ADA 4H, well above walk-forward threshold without rare-event risk.

Implements a NEW single-bar mechanism for ADA 4H: raw range PERCENTILE (top quartile over 50 bars) combined with close-within-range geometry, deliberately distinct from Donchian channels, smoothed-ATR expansion, multi-bar close-position averaging, return z-scores, channel-width expansion, and sequence patterns. The conjunction provides built-in noise filtering -- a wide-range bar that closed near its low (sellers won) is rejected, while a wide-range bar that closed near its high (buyers absorbed the flow) triggers, capturing momentum-injection / accumulation events. calculate_signal returns a continuous, bar-varying score (range ratio x close position) so the verifier sees real signal variation, while the discrete percentile/close-position trigger gates should_enter and position_size, avoiding the gated-to-0.0 anti-pattern. Exits use an ATR stop/target plus a 6% fallback and a 12-bar time stop to ride the injection without overstaying. Venue ADAUSDT.BINANCE USD-M 4H is the most-reliably-fresh ADA bar type (avoids the audit_stale pattern); long-only with leverage left at 1.0 (no amplification used) so the leverage_set_but_unused gate does not trigger, and the 2.5x ATR / 6% targets keep expected per-trade edge above the ~0.10% round-trip fee. Single instrument / single timeframe avoids cross-asset collection failures.

ADA single-bar range-outlier + bullish-close is an essentially break-even, untradeable signal: profit_factor 1.037, Sharpe 0.146 (CI straddles 0), and per-trade expectancy ~0.12% which is BELOW the ~0.15% futures cost floor. Critically it is untradeable at scale — impact_cost_pct 35.6% (impact eats over a third of gross PnL) with capacity only ~$790K — so the apparent edge exists only at near-zero size. The signal also over-fires: 710 trades (~118/yr) vs the hypothesis's projected ~35-55/yr, generating $17k commission + $8.6k impact churn with no per-trade edge to cover it, and the result is front-loaded into 2021 (+17.8%) and 2023 (+13.1%) with all three recent years negative (2024 -2.9%, 2025 -5.6%, 2026 -5.7%, rolling Sharpe ~-3 by 2026). The code is correct and trades as specified, so this is no-edge economics plus a structural liquidity limit, not a code defect — iterate cannot help, and optimizing a PF-1.037 base would fit the 2021/2023 bull and fail the recent holdout. Not worth 2 hours. This is even weaker than the LINK/AVAX Keltner siblings already abandoned. FAILURE PATTERN: single-bar range-outlier + bullish-close momentum-capture on ADA 4H produces a break-even, untradeable result (PF 1.04, Sharpe 0.15, per-trade below the cost floor, impact 35.6%, capacity ~$790K) — the conjunction fires too often to be selective, so it churns fees with no fee-surviving edge and is front-loaded/decaying; single-bar geometric breakout signals on a lower-liquidity alt do not generate a deployable edge.