BTC Daily Volatility-Compression Expansion Long with 200-SMA Regime Filter

A long-only single-instrument volatility-regime strategy on BTCUSDT perpetual futures using daily bars and OHLCV-only data. This is a STRUCTURALLY NEW mechanism not present in any existing pipeline strategy: it captures the empirical pattern that low-volatility compression in BTC tends to be followed by directional expansion, and during confirmed bull regimes the expansion is biased UPWARD. The strategy enters long when BTC is in a vol-compression regime (realized volatility in bottom 30% of trailing 90-day distribution) AND the long-term regime is up (price > 200-SMA), then holds until volatility normalizes (returns to 50th percentile) or the regime breaks. This explicitly differs from all 12 existing pipeline strategies: NOT a trend-following strategy (no EMA crossovers, no breakouts, no momentum filters), NOT a drawdown DCA (no buy-the-dip mechanics), NOT a volume breakout (no volume signal), NOT a regime-flat hold (uses vol regime, not just price regime). The mechanism is 'Bollinger squeeze' style — buy compression, exit on expansion completion. Critically, this proposal addresses the AVAX Calm-Regime Trend Hold failure (abandoned for sparsity + outlier dependency) by: (1) selecting BTC instead of AVAX — BTC has 3-5× the frequency of vol-compression regimes due to its lower baseline volatility and deeper liquidity, (2) using a single dominant filter (vol percentile) instead of triple-AND-gate (low vol + uptrend + rising SMA + EMA stack), (3) targeting 8-15 entries/year (well above the 2.5/year sparsity threshold). BTC-SPECIFIC empirical justification: BTC has multiple WORKING strategies confirming clean vol/price dynamics (BTC LT Momentum Sharpe 5.91, BTC Spot Drawdown Sharpe 3.33, BTC 4H Volume Breakout Sharpe 3.74). The pattern of low-vol-then-expansion is documented in TradFi (Bollinger, Connors) and applies cleanly to BTC's institutional-flow-dominated microstructure.

Implements the Bollinger/Connors squeeze mechanism (buy compression, exit on expansion completion), a structurally new vol-regime edge distinct from trend-following, drawdown DCA, and volume breakouts. Volatility is measured as a self-normalizing percentile within its own trailing distribution, so 'bottom 30%' adapts across BTC's changing vol regimes rather than depending on an absolute threshold. calculate_signal returns the continuous, bar-varying compression strength so the verifier sees real signal variation, while the 200-SMA bull-regime check is stored as self._regime_ok and ANDed in should_enter/position_size against the entry percentile, avoiding the gated-to-0.0 anti-pattern. A single dominant filter (vol percentile) plus one regime gate keeps the trigger frequency high (targeting 8-15 entries/year on BTC), addressing the prior AVAX calm-regime sparsity failure by choosing BTC (3-5x more compression regimes) and dropping the multi-AND gate. Exit on vol normalization captures the expansion move and steps aside once the edge is gone, with a hard stop and time stop only as tail protection. min_bars_required clears 200+ so SMA200 and the 90-vol distribution are fully formed before trading. Venue BINANCE USD-M futures matches the clean, proven BTC data family and futures RT taker (~0.10%) is cheaper than spot; long-only with leverage left at 1.0 (no amplification) so the leverage_set_but_unused gate does not trigger.

BTC daily vol-compression->expansion has NO realized edge: profit_factor 0.842 (<1), negative expectancy (-$673.7/trade), Sharpe 0.086 with a CI straddling 0. The +6.9% headline is illusory — end_unrealized_pct 23.1 exceeds total_return 6.9, so the realized closed-trade record is negative (~-16%); the headline is an un-exited open position marked-to-market on the last bar. The only positive contribution is a single +53.8% day (2024-02-28, kurtosis 141.8, skew 7.26, tail_ratio 0.0057), alongside a catastrophic 44.8% max drawdown. The strategy also reproduced the exact failure it claimed to avoid: it targeted 8-15 entries/year but produced only 24 trades over 6.5 years (~3.7/yr), sparse and outlier-dependent like the cited AVAX Calm-Regime abandonment. The code is correct and trades as specified, so this is a premise failure, not a bug — BTC vol-compression produces a negative realized expectancy that optimization cannot fix, and the sparse 24-trade sample would only let the optimizer fit the lone 2024 outlier and fail the holdout. Not worth 2 hours. FAILURE PATTERN: single-asset daily volatility-compression ('Bollinger squeeze') long on a crypto major produces a sparse (~3.7/yr), outlier-dependent, net-losing realized record (PF 0.84, negative expectancy) whose positive headline is entirely open-position unrealized MTM plus one +53.8% day — a vol-regime entry filter on a major does not generate a fee-surviving directional edge and cannot be validated given the event sparsity and 45% drawdown.