AdaFourHourDonchianMidpointReclaimLong

ADA 4H Donchian Channel-Midpoint Reclaim Long (Structural Mid-Range Recovery)

A long-only, single-instrument, single-timeframe strategy on ADAUSDT.BINANCE 4H bars (audit_stale issue now RESOLVED per analyst feedback). Implements a NEW STRUCTURAL MECHANISM consistent with the analyst's explicit guidance: 'STRUCTURAL signals (sequence-based or range-statistics-based) are what work on liquid crypto.' The mechanism: DONCHIAN CHANNEL-MIDPOINT RECLAIM. Compute the 20-bar Donchian channel midpoint (average of high20 and low20). When price has been BELOW the midpoint for at least 3 consecutive bars (price was in the lower half of the range — sellers controlled the range) and then the current bar CLOSES BACK ABOVE the midpoint by at least 0.2×ATR (decisive reclaim, not noise), this signals a structural shift back to buyer control within the established range. Combined with a long-term uptrend filter (close > sma_100), this captures the moment when price recovery within range aligns with broader trend. Critically distinct from every existing ADA 4H mechanism: (a) NOT Donchian high/low breach (uses MIDPOINT, not channel boundary); (b) NOT double-low support (uses INTRA-RANGE midpoint pivot, not low-level matching); (c) NOT failed-breakdown reclaim (different reference: midpoint not low); (d) NOT channel-WIDTH expansion (uses midpoint as price reference, not range size); (e) NOT cumulative return / z-score / autocorrelation (no statistical-magnitude math — pure structural price-level signal); (f) NOT volume signal. STRUCTURAL + RANGE-STATISTICS-BASED per analyst's explicit recommendation. Three filters only.

Implements the analyst-recommended structural / range-statistics mechanism using the Donchian MIDPOINT as the reference pivot -- explicitly distinct from boundary-breach Donchian (channel high/low), double-low support, failed-breakdown reclaim, channel-width expansion, and any statistical-magnitude math. The midpoint is the range's control level: a >=3-bar run of closes below it means sellers owned the range, and a decisive >=0.2xATR reclaim flips control to buyers within the range, filtered by an SMA100 uptrend so the intra-range recovery aligns with the broader trend. calculate_signal returns the continuous, bar-varying midpoint distance in ATR units so the verifier sees real signal variation, while the discrete reclaim boolean gates should_enter and position_size, avoiding the gated-to-0.0 anti-pattern; the below-midpoint streak is maintained as running state and read as the PRIOR run so the reclaim bar itself is not miscounted. Exits use an ATR stop/target, a midpoint-loss structural stop, a percentage fallback, and a time stop. Single instrument / single timeframe avoids cross-asset audit_stale risk (now resolved for ADA 4H). Venue ADAUSDT.BINANCE USD-M futures matches the proven ADA 4H family and futures RT taker (~0.10%) is cheaper than spot; long-only with leverage left at 1.0 (no amplification used) so the leverage_set_but_unused gate does not trigger, and the ATR/6% targets keep expected per-trade edge above the ~0.10% round-trip fee.

Marginal, capacity-constrained, decaying edge that is not worth optimizing. The strategy is implemented correctly and net-positive (133 reclaim entries, +5.8% over 6.4 years, impact folded), but the edge is razor-thin: profit_factor 1.067, Sharpe 0.227 (CI straddles 0), win_rate 30%, and ~0.14% net per trade — at/below the BINANCE USD-M cost floor. Critically, impact_cost_pct is 28.9% with capacity only ~$1.2M, so the apparent edge is largely a small-size artifact that vanishes at any deployable scale (a promotion blocker on its own), and the strategy is decaying out-of-sample: 2024 +5.8% fades to 2025 -4.3% and 2026 -5.3% with the rolling Sharpe collapsing to -10 by 2026 (max_consecutive_losses 14). This is an economics/premise problem, not a code defect, so iterate cannot help; and optimizing a PF-1.067, decaying base would at best fit the 2020-2024 period and then fail deflated Sharpe and the recent (2025/2026) holdout. Not worth 2 hours. FAILURE PATTERN: marginal long-only structural reclaim/reversion strategies on ADA 4H produce a thin net edge (PF ~1.07, Sharpe ~0.2, ~0.14% per trade) that is real only at toy scale (impact 28.9%, capacity ~$1.2M) and decays out-of-sample (2025/2026 negative, rolling Sharpe -10) — the same marginal-edge, recent-decay signature as the abandoned DOT volume-breakout and SOL reversal siblings, which optimize onto the historical bull years and then fail DSR and the recent holdout.

Long-only Donchian channel-midpoint reclaim on ADAUSDT.BINANCE USD-M futures, 4H bars, pure OHLCV single feed. Computes the 20-bar Donchian midpoint ((high20+low20)/2); enters LONG when price has closed below the midpoint for >=3 consecutive bars and then closes back above it by >=0.2x ATR (decisive reclaim) while in a long-term uptrend (close > SMA100). Exits on a 1.5x ATR stop, a 3x ATR target, a 6% take-profit fallback, a midpoint loss (close back below the midpoint), or a 30-bar time stop. The signal is the continuous midpoint distance (close-midpoint)/ATR recomputed every bar; the discrete reclaim is the entry gate. Sized at 18% equity notional, capped so ATR-stop risk <= 1.5% of equity. leverage 1.0.

Implemented correctly and trades as specified: 133 midpoint-reclaim entries, 0 dropped, all long, exposure 7.1%; genuine structural mechanism with a healthy sample

Net-positive overall (+5.8%) with low drawdown (10.2%) and positive in 5 of 7 years

Razor-thin edge: profit_factor 1.067, Sharpe 0.227 (CI -1.69 to 1.89), win_rate 30%, expectancy +$25/trade (~0.14% net) — at/below the BINANCE USD-M viability floor

Capacity-constrained, impact-fragile: impact_cost_pct 28.9% with capacity only ~$1.2M — the edge is real only at toy scale, a promotion blocker on its own

Recent decay: annual returns fade 2024 +5.8% -> 2025 -4.3% -> 2026 -5.3%, rolling Sharpe to -10 by 2026, max_consecutive_losses 14 — the holdout (recent window) would likely fail

A PF-1.067 / decaying base will, at best, optimize to fit 2020-2024 and fail deflated Sharpe + the recent holdout (the DOT/SOL marginal-edge pattern)

This strategy introduced a structural Donchian midpoint-reclaim mechanism on ADA 4H — buying decisive recoveries back above the range midpoint after a 3-bar below-streak in an uptrend — implemented cleanly over 133 trades and net-positive (+5.8%, 10.2% drawdown). But the edge was razor-thin (profit factor 1.067, Sharpe 0.227, ~0.14%/trade), real only at toy scale (28.9% impact cost, $1.2M capacity), and decaying with 2025 and 2026 negative and rolling Sharpe near -10. The analyst abandoned it at the backtest-review gate on the first iteration as a marginal, capacity-constrained, decaying edge not worth optimizing — matching the DOT/SOL marginal-edge pattern — so it never advanced to optimization or risk review.

A correctly-implemented structural signal with a positive headline can still be uninvestable — a profit factor near 1.07 at ~0.14%/trade is below the cost floor, exists only at ~$1.2M toy scale given 28.9% impact, and decays out-of-sample, so it would optimize onto historical bull years and fail forward validation.

The backtest-review analyst issued an 'abandon' verdict: the code was correct and net-positive, but the edge is economically marginal (PF 1.067, ~0.14%/trade), real only at toy scale (28.9% impact, $1.2M capacity), and decaying out-of-sample — so optimization would at best fit the 2020-2024 bull years and then fail deflated Sharpe and the recent holdout, the same marginal-edge/recent-decay signature as the abandoned DOT and SOL siblings.

A long-only structural mean-reversion strategy on ADAUSDT 4H Binance futures (1.0x leverage, 0.18 position) — entering long when price, after closing below the 20-bar Donchian channel midpoint for 3+ consecutive bars, decisively reclaimed it (closing ≥0.2 ATR above the midpoint) while above a 100-SMA uptrend filter — a midpoint-reclaim pivot distinct from boundary-breach, double-low, and width-expansion mechanisms.

Across 133 trades it was net-positive but razor-thin: +5.8% total over 6.4 years, profit factor 1.067, Sharpe 0.227 (CI straddling zero), a 30.1% win rate, expectancy +$25/trade (~0.14%, at/below the cost floor), and a low 10.2% drawdown; capacity was only ~$1.2M with 28.9% impact cost, and returns decayed from 2024 +5.8% to 2025 -4.3% and 2026 -5.3% with rolling Sharpe collapsing toward -10.