SpotIntradayVolumeSpikeCapitulationFlushReversalBasketLong

Intraday Volume-Spike Capitulation-Flush Reversal Spot Basket, Long-Only (3 Liquid BINANCE_SPOT Alts: SOL/AVAX/LTC, 1h Bars, Buy Only Single-Bar Forced-Selling Flushes Confirmed by an Abnormal Volume Z-Spike, Exit on Partial Reversion or Fast Time-Stop, 3-Parameter)

A LONG-ONLY intraday mean-reversion basket on three liquid Binance SPOT alts (SOLUSDT, AVAXUSDT, LTCUSDT) that trades a specific, crypto-native microstructure event: the single-bar CAPITULATION FLUSH. When a stop cascade or forced-selling wave hits, price gaps down intrabar on ABNORMALLY high volume and typically overshoots fair value, then partially V-recovers within hours as passive dip-buyers and liquidated shorts re-cover. The edge is deliberately NOT a slow multi-day drawdown (that is the separate SpotUptrendGatedDeepDipReversionBasket / LinkSpot4H strategies already in pipeline) and NOT a static price-level dip — the entry is EVENT-gated by a VOLUME Z-SPIKE, the microstructure twist that distinguishes a genuine forced-selling flush from ordinary drift. Each name is evaluated independently on 1h bars. Routed to BINANCE_SPOT (under-quota at 13.0%, the factory's highest-survival venue), long-only (CASH account, no shorting needed), and kept to 3 parameters to resist the overfitting that is the #1 killer (239x). Small fixed basket avoids the equity-accounting bugs that killed the 8-name breadth and ETH/BTC baskets while still giving enough independent events for a measurable sample.

Implements the event-gated capitulation-flush edge: the entry fires only on the conjunction of a sharp single-bar drop AND an abnormal volume z-spike, which is the microstructure signature that distinguishes forced-selling overshoot (which V-recovers) from ordinary drift (which does not) — deliberately distinct from the slow multi-day-drawdown dip strategies already in pipeline. Kept to 3 tuned parameters (the two that define the event, vol_z_thresh and flush_ret, plus the reversion exit fraction) to resist the overfitting that is the dominant killer; the volume baseline, time-stop, protective stop, and gross are fixed structural choices. Routed to BINANCE_SPOT (CASH, long-only, leverage 1.0) because the edge is an unleveraged intraday accumulation needing no shorts/funding, and all three USDT-quoted legs are liquid with long spot histories (the account funds USDT so every leg is directly buyable — no BTC-quote or data wall). A small 3-name basket gives enough independent flush events for a measurable sample on 1h bars while avoiding the equity-accounting fragility of the larger baskets. Execution reuses the deployed multi-instrument machinery: own symbol-keyed close/volume buffers updated from every bar, an on_bar override, and true-NAV _portfolio_value marking so equal-weight sizing keeps gross <= 0.95x with no negative-cash/over-deploy path; but unlike the daily rotation siblings it evaluates each name on ITS OWN freshly-arrived 1h bar (independent intraday events — no cross-leg alignment lag) with a per-symbol monotonic bar counter as the time-stop clock. Fee headroom: a 3% flush recovering half is a ~1.5% move, ~7x the 0.20% spot round-trip, and holds are hours so turnover stays modest. calculate_signal returns a continuous flush score for the Layer-2 proxy while the real allocation runs in _evaluate_symbol; should_enter/should_exit/position_size are inert stubs as in the proven multi-instrument pattern." }

No edge to optimize: on 372 well-sampled trades the strategy is a net loser — total_return -18.4%, profit_factor 0.931, expectancy -$49.5/trade, Sharpe 0.019 (CI straddles zero). The failure is diagnostic: win_rate 60.5% but avg_loss $1,824 ≈ 1.64x avg_win $1,110 — the reversion take-profit banks tiny V-recoveries while the 5% stop lets cascading flushes run, so it wins often and loses big, netting negative (the same inverted reward/risk that has killed the intraday reversion-fade class repeatedly this session). avg_trade_return_pct 0.268% only grazes the 0.25% spot fee floor and the dollar-weighted result is negative. Compounding this, capacity_usd is $274 with impact_cost_pct 1910% — market impact eats ~19x gross PnL, so even a real edge would be uninvestable beyond toy scale. Returns are regime-fragile (2022 -26%, 2024 -8%, 2025 -15%, carried by one 2021 month). Optimizing the 3 params cannot flip negative expectancy or the inverted RR, and no reframe of intraday capitulation-flush buying clears both the fee floor and the impact wall. Abandon rather than spend 2 hours optimizing.

Long-only intraday capitulation-flush reversal basket on three liquid Binance SPOT alts (SOL/AVAX/LTC), 1h bars, pure OHLCV. Each name independently: when a single 1h bar drops >= flush_ret (3%) below the prior close AND that bar's volume is a >= vol_z_thresh (3.0) z-score spike above its prior 48-bar baseline (a genuine forced-selling flush, not drift), buy an equal-weight slice. Exit on partial reversion (price recovers reversion_frac=0.5 of the flush back toward the pre-flush level), a protective stop (5% below entry if the cascade continues), or a fast 12-hour time-stop. Slices are sized off true CASH portfolio equity (USDT cash + holdings mark) at gross_exposure/3 each, so gross never exceeds 0.95x (no leverage, spot long-only). 3 tuned params (vol_z_thresh, flush_ret, reversion_frac).

Analyst: point estimate is a clear loser (PF 0.438) though the Sharpe CI straddles zero on 38 trades — abandon on the inverted RR, or if optimizing, the burden is to show a regime (stricter vol_z_thresh/deeper flush_ret, or an asymmetric exit letting winners run past the half-reversion target) that flips avg_trade_return_pct positive OOS. Gate loosening won't help — losers being ~2x winners is the structural problem.

Negative edge with an INVERTED reward/risk — analyst should scrutinize heavily and most likely ABANDON. On 38 well-sampled trades (metrics_reliable=TRUE): profit_factor 0.438, total_return -12.9%, avg_trade_return_pct -0.803%, Sharpe -1.11 (CI [-2.57, +0.76]). win_rate 0.447 but avg_loss 1092 ≈ 1.85x avg_win 590 (skew -3.9, largest_loss -2446 vs largest_win 1292): the flush too often keeps cascading into the 5% stop rather than V-recovering — wins often, loses big, the classic mean-reversion-fade inverted-RR signature. Genuine directional failure, NOT a fee problem: -0.803%/trade is ~4x the ~0.20% spot round-trip, turnover modest (16.6).

Well-sampled (372 trades, all long as the spot design requires) and runs cleanly with no accounting/liquidation bug (end_unrealized 0%).

Net loser: total_return -18.4%, profit_factor 0.931, expectancy -$49.5/trade, Sharpe 0.019 (CI [-0.67, 0.68] straddles zero).

Inverted reward/risk despite 60.5% win rate: avg_loss $1,824 is ~1.64x avg_win $1,110 — small reversion take-profits vs large stop losses when the flush keeps cascading.

avg_trade_return_pct 0.268% barely at the 0.25% spot floor and dollar-weighted result is negative — no fee-clearing edge.

Catastrophic capacity: capacity_usd $274, impact_cost_pct 1910% — modeled impact ~19x gross PnL; uninvestable beyond toy scale.

Regime-fragile: 2022 -26%, 2024 -8%, 2025 -15%; carried only by a single 2021 month.

Intraday reversion-fade class that has died repeatedly with this exact inverted-RR signature.

SpotIntradayVolumeSpikeCapitulationFlushReversalBasketLong tried to buy volume-confirmed single-bar capitulation flushes across a small SOL/AVAX/LTC spot basket, expecting partial V-recoveries within hours. It executed cleanly with no accounting bug over 372 well-sampled long trades, but was a net loser — total_return -18.4%, profit factor 0.931, -$49.5 expectancy, Sharpe ~0.02 — carried down by the classic inverted reward/risk of the intraday reversion-fade class: it won 60.5% of the time but lost ~1.64x more on the losers when flushes kept cascading. A $274 capacity with 1910% impact cost (≈19x gross PnL) and regime fragility (2022 -26%, 2024 -8%, 2025 -15%, carried by one 2021 month) confirmed there was nothing to optimize. The analyst abandoned it at the pre-optimization gate, and optimization, risk review, and later stages were never reached.

A high win rate does not imply an edge when reward/risk is inverted — buying intraday capitulation flushes banks tiny V-recoveries while stops let cascades run to large losses, and the resulting sub-fee, tiny-capacity result cannot be rescued by parameter tuning.

The backtest-review analyst issued an abandon verdict before optimization: there was no edge to tune — negative expectancy driven by the inverted reward/risk signature (small reversion take-profits vs large stops when flushes kept cascading), a fee-grazing per-trade return, regime fragility, and a catastrophic capacity/impact wall (~19x gross PnL) that made it uninvestable beyond toy scale.

A long-only intraday mean-reversion basket on three liquid BINANCE_SPOT alts (SOL/AVAX/LTC, 1h bars) that bought single-bar capitulation flushes confirmed by an abnormal volume z-spike, exiting on partial V-reversion, a protective stop, or a fast time-stop.

Across a well-sampled 372 long trades it was a net loser: total_return -18.4%, profit factor 0.931, expectancy -$49.5/trade, and Sharpe 0.019 with CI [-0.67, 0.68] straddling zero. Despite a 60.5% win rate the reward/risk was inverted (avg_loss $1,824 ≈ 1.64x avg_win $1,110), avg_trade_return_pct 0.268% only grazed the 0.25% spot fee floor, and capacity was just $274 with impact_cost_pct 1910%.