BtcSpotVolumeSpikeCapitulationFlushBounceLong

BTC Spot Intraday Volume-Spike Capitulation-Flush Bounce (Long-Only, BINANCE_SPOT, 1H, OHLCV Proxy for Forced-Liquidation Flow)

A LONG-ONLY, single-instrument, pure-OHLCV intraday strategy on BTCUSDT.BINANCE_SPOT using 1-HOUR bars that BUYS capitulation flushes — sharp down-moves on abnormal volume spikes — to capture the mechanical bounce. This is the analyst's explicitly-recommended pivot: after FOUR consecutive supplementary-stream-primary abandons this session (open_interest x2, account-ratio, liquidations — all only ~weeks of catalog history), the analyst recommended replacing the liquidation-cascade-reversal edge with an 'OHLCV-derivable proxy for forced flow (large-range-bar reversal, volume-spike reversal).' A liquidation cascade leaves its fingerprint directly in OHLCV: an outsized down-candle on an abnormal volume spike IS a forced-liquidation flush. EVERY design choice is anchored to a thing that has actually WORKED or a documented kill: (1) PURE OHLCV (volume-spike + range) — NO supplementary data, immune to the open_interest/taker/L-S/liquidations data walls that killed every primary-supplementary strategy. (2) LONG CAPITULATION-BOUNCE on BTC — the same proven-direction, proven-asset edge as the PROMOTED BTC-daily drawdown-accumulation (buy forced down-moves on low-vol BTC where reversion physics holds), NOT the short-into-squeeze direction that killed the XRP OI-fade, and NOT the high-vol-alt fade that killed the SOL VWAP-MR. (3) INTRADAY frequency — the daily drawdown-accumulation TRANSFERS died of event-sparsity (deep multi-day drawdowns too rare to populate OOS windows); an intraday 1H volume-spike flush fires several times per month, so every walk-forward OOS window is densely populated. (4) VOLUME-SPIKE GATE = the OHLCV signature of MECHANICAL forced flow, distinguishing a margin-call flush (overshoots, bounces) from an informational down-trend (continues) — the filter that makes this fade survive where naked excursion-fading failed. (5) LONG-ONLY routes correctly to BINANCE_SPOT (CASH, no leverage), filling the under-target spot venue (11.5% vs 15% target). (6) NOT a closed family: NOT a funding/OI/liquidations primary-signal strategy, NOT a funding-confirmed-trend clone (which failed to transfer off ETH), NOT cross-asset ratio, NOT breakout/momentum, NOT COIN-M/HL. Intentionally minimal: 3 tunables (flush_ret_pct, vol_spike_mult, max_hold_bars).

Implements the analyst's pivot away from supplementary-stream-primary strategies (four data-wall abandons) by deriving the forced-flow signal directly from OHLCV: an outsized down-candle on an abnormal volume spike IS a liquidation flush. Every design choice is anchored: (1) PURE OHLCV (volume-spike + range) so it is immune to the open_interest/taker/L-S/liquidation coverage walls; the volume baseline is a trailing median, robust to outliers. (2) LONG capitulation-bounce on BTC — the same proven direction/asset as the promoted BTC drawdown-accumulation (buy forced down-moves on low-vol BTC where reversion holds), NOT the short-into-squeeze that killed the XRP OI-fade nor the high-vol-alt fade that killed the SOL VWAP-MR. (3) INTRADAY 1H frequency so flushes fire several times per month and populate every OOS window, avoiding the event-sparsity that killed the daily drawdown transfers. (4) The VOLUME-SPIKE GATE is the crux — it distinguishes a mechanical margin-call flush (overshoots, bounces) from an informational down-trend (continues), the filter that lets this fade survive where naked excursion-fading failed; the stop-loss cleanly cuts the cases where it was informational. (5) LONG-ONLY routes correctly to BINANCE_SPOT (CASH, leverage=1, sizing uses no leverage so leverage=1.0 is correct and not flagged), filling the under-target spot venue; entries are BUY-only and exits sell the long via the base helper (no OrderSide.SELL-to-open). calculate_signal returns a CONTINUOUS graded flush score (-ret x clamped volume ratio) that varies every bar and carries conviction (defeats the Layer-2 frozen-signal check), while the discrete two-condition entry lives in should_enter. Kept to exactly 3 tunables. Not a funding/OI/liquidations-primary, funding-confirmed-trend, cross-asset-ratio, breakout/momentum, or COIN-M/HL family.

Fundamentally negative-edge mechanism, not a data or tuning problem. The BTC spot 1H volume-spike capitulation-flush bounce loses money in 7 of 9 years (2018-2026; only 2023 +0.05% and 2026 +1.2% are non-negative), with profit_factor 0.74, Sharpe -2.36 (CI entirely negative), expectancy -$86/trade, -27.6% total return, and probabilistic_sharpe 0.0016 over a clean 374-trade pure-OHLCV sample. Two structural problems no parameter sweep can fix: (1) WIN/LOSS INVERSION + WRONG EXIT ASYMMETRY - win_rate 51.3% but avg_loss ($668) is 1.4x avg_win ($466), because the take-profit (1.5%) is smaller than the stop (2.0%) AND the volume-spike down-flush CONTINUES (informational down-move) often enough that the spike gate fails to separate mechanical forced flow from genuine downtrends; buying the flush is a fade, and the flush trends through the bounce. (2) FEE DOMINANCE - commission is 21.9% of gross PnL at 1H frequency with a small 1.5% target, so the strategy is fee-fragile even before the negative gross edge. This is the same intraday-fade failure mode as this session's VWAP-excursion family (which lost every year on BTC/SOL): fading/bouncing intraday down-moves on crypto majors is a structural anti-edge because the moves continue more than they revert. Recommend abandon rather than spend 2 hours optimizing a 7-of-9-year loser with 22% fee drag. FAILURE PATTERN: long-only intraday volume-spike capitulation-bounce on a crypto major is a net-negative fade - the OHLCV volume-spike gate does not reliably distinguish mechanical flushes (that bounce) from informational selloffs (that continue), so the payoff inverts (avg_loss > avg_win) and 1H fee drag (~22% of gross) erases any residual edge; intraday buy-the-flush on majors should not be re-proposed.

Long-only single-instrument intraday capitulation-bounce on BTCUSDT.BINANCE_SPOT (CASH, no leverage), 1-HOUR pure-OHLCV bars. Each bar it measures the candle return (close/prev_close - 1) and the volume ratio (volume / median volume over vol_lookback=48 prior bars). It BUYS when BOTH a sharp down-move (ret <= -flush_ret_pct, 2%) AND an abnormal volume spike (vol_ratio >= vol_spike_mult, 2.5x) occur — the OHLCV signature of a forced-liquidation flush that mechanically overshoots and bounces. It exits on the earliest of a 1.5% take-profit, a 2% stop-loss (the flush kept going = informational down-trend, cut), or a 12-bar (~12h) time stop. Three tunables (flush_ret_pct, vol_spike_mult, max_hold_bars). Long-only (BUY to enter, sell to exit), leverage=1.

Clean execution and data: pure OHLCV (no data wall), full 2018-2026 coverage, 374 trades, no liquidation, correct spot routing.

Selective (exposure 3.6%) and well-motivated as an OHLCV proxy for forced-liquidation flow.

Negative in 7 of 9 years (2023/2026 only marginally positive); total_return -27.6%, Sharpe -2.36 (CI entirely negative), probabilistic_sharpe 0.0016.

profit_factor 0.74, expectancy -$86/trade, max_drawdown 31% with a 2846-day (never-recovered) drawdown.

Win/loss inversion: avg_loss $668 is 1.4x avg_win $466 — the TP 1.5% < SL 2.0% design caps winners while flushes that continue run to the bigger stop.

Fee-dominated: commission is 21.9% of gross at 1H — a small take-profit at intraday frequency is structurally fee-fragile.

Same intraday-fade failure mode as this session's VWAP-excursion family: volume-spike down-flushes continue more than they bounce.

BtcSpotVolumeSpikeCapitulationFlushBounceLong was the analyst's recommended escape from the data-blocked supplementary-stream strategies — using a volume-spike-plus-down-candle as a pure-OHLCV fingerprint of forced-liquidation flushes, on the proven-direction BTC-long-capitulation premise, at intraday frequency to beat event-sparsity. The data was clean and the design selective, but the edge was negative: across 374 trades it lost -27.6% with a -2.36 Sharpe and 0.74 profit factor, negative in seven of nine years, because flushes continue more than they bounce and a TP<SL design plus 22% fee drag inverted the payoff. The analyst ruled it the same intraday-fade anti-edge as the session's VWAP-excursion family and abandoned it, advising against re-proposing intraday buy-the-flush on majors. It ended after one iteration as abandoned, never advancing to optimization or risk review.

Long-only intraday buy-the-flush (volume-spike capitulation bounce) on a crypto major is a net-negative fade: an OHLCV volume-spike gate does not reliably distinguish mechanical flushes that bounce from informational selloffs that continue, so the payoff inverts (avg_loss > avg_win) and 1H fee drag erases any residual edge — the same intraday-fade anti-edge as the VWAP-excursion family, so buying intraday down-moves on majors should not be re-proposed.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) as a structural negative-edge fade — volume-spike down-flushes continue more than they bounce, so the spike gate failed to separate mechanical flushes from informational selloffs, and the TP(1.5%) < SL(2.0%) asymmetry plus ~22% 1H fee drag inverted the payoff — making it unfixable by tuning, so optimization and all later stages were never reached.

A long-only single-instrument intraday capitulation-bounce on BTCUSDT.BINANCE_SPOT (1H bars, OHLCV-only, CASH/no leverage) that bought sharp down-candles on abnormal volume spikes — entering when the bar return fell <=-2% and volume exceeded 2.5x its 48-bar median, exiting on a 1.5% take-profit, a 2% stop, or a 12-bar time stop — using the volume-spike-plus-range signature as an OHLCV proxy for forced-liquidation flushes, the analyst's recommended pivot after four supplementary-stream-primary abandons.

Over 2018-2026 and a clean 374-trade pure-OHLCV sample (no liquidation, very selective at 3.6% exposure) it was a negative edge: total return -27.6%, Sharpe -2.36 (CI entirely negative), profit factor 0.74, expectancy -$86/trade, max drawdown 31% (2,846-day never-recovered), PSR 0.0016, negative in 7 of 9 years, with a win/loss inversion (51.3% win rate but avg loss $668 = 1.4x avg win $466) and fees consuming 21.9% of gross.