SOL Intraday Daily-Anchored-VWAP Excursion Mean-Reversion (Single-Instrument USD-M, Long-Short, 1H, Continuously-Triggering, VWAP-Touch Exit, ATR-Stopped)

A LONG-SHORT, SINGLE-INSTRUMENT, pure-OHLCV intraday MEAN-REVERSION strategy on SOLUSDT.BINANCE USD-M perpetual futures using 1-HOUR bars. It is a deliberate sibling of the factory's PROMOTED ETH 1H daily-anchored-VWAP mean-reversion strategy, replicated on a higher-volatility asset, and engineered to AVOID every documented failure mode of that mechanism's abandoned long-only variant. Mechanism: compute the daily-anchored VWAP (volume-weighted average price reset at 00:00 UTC, accumulated from the day's 1H bars). Price wanders above/below this intraday fair-value anchor and tends to revert to it within the session. When price is significantly BELOW VWAP, go LONG (expect reversion up); when significantly ABOVE, go SHORT (expect reversion down). CRITICAL DESIGN CHOICES, each correcting a specific prior failure: (1) PURE OHLCV — uses NO supplementary data, completely avoiding the OI/taker/funding multi-year coverage holes that abandoned five strategies this session (OI-surge, taker-flow, XRP-OI-fade, etc.); VWAP is computed from the 1H bars themselves, which have full multi-year coverage. (2) LONG-SHORT, NOT long-only — the abandoned ETH variant was long-only-bull-regime-gated, which structurally confined trades to bull markets and starved OOS windows; this fades both sides in every regime. (3) NO regime gate and a MODEST, vol-normalized excursion threshold (~1.8 sigma, NOT a rare 3%+ deep excursion) chosen so entries fire CONTINUOUSLY — explicitly targeting >=10 qualifying entries per walk-forward OOS window, directly defeating the event-sparsity / OOS-all-zero / IS-Sharpe-11 trade-starvation signature that killed the long-only variant and the drawdown-accumulation family. (4) ATR STOP to contain tail squeezes — the XRP fade blew up on an uncontained -29.9% SEC-ruling squeeze; a tight ATR stop plus fading only MODEST (not violent) excursions limits the left tail. NOT a closed family: NOT cross-asset pairs-ratio MR (single instrument vs its own intraday anchor, not a non-stationary 2-asset ratio), NOT breakout/momentum, NOT cross-venue/HL, NOT COIN-M. Intentionally minimal: 3 tunables (entry_z, vol_lookback, atr_stop_mult).

Single-instrument symmetric intraday mean-reversion using the standard FactoryStrategy hooks, replicating the promoted ETH 1H daily-anchored-VWAP mechanism on higher-vol SOL while correcting each documented failure of the abandoned long-only variant. (1) Pure OHLCV: the VWAP and z-score are computed only from the 1H bars (full multi-year coverage), with no OI/taker/funding supplementary data, avoiding the coverage-hole abandons. (2) Long-short with no regime gate: it fades both sides in every regime, so entries are not confined to bull markets. (3) Continuous triggering: the z-score normalizes (close-VWAP) by a rolling vol_lookback-bar std, so a modest ~1.8-sigma threshold fires densely across regimes (targeting >=10 entries per OOS window) rather than a rare deep-excursion event -- defeating the trade-starvation/OOS-all-zero signature. (4) Left-tail control: it fades only modest excursions and exits on an ATR(14) 2.5x stop, containing the uncontained-squeeze blowups that killed the XRP fade. The daily VWAP anchor resets at UTC midnight via integer ns math (day = ts_ns//NS_PER_DAY) and time gates use hour = (ts_ns//NS_PER_HOUR)%24 (no datetime import; epoch ns is UTC); a 23:00 forced flat ensures no position is held across the anchor reset (stale-anchor risk), and entries stop at 22:00 so positions have time to revert. calculate_signal returns the continuous z-score (varies every bar) so Layer-2 frozen-signal detection passes, while the discrete entry gates (z thresholds, MIN_DAY_BARS=4 warmup) live in should_enter. Robustness: a zero-volume bar falls back to the mean of the day's closes so VWAP stays defined, std<=0 yields z=0 (no entry), and the side/entry context is reconstructed from avg_px_open if lost. Sizing is capital-relative (10% of equity) clearing the SOLUSDT $5 min-notional; the ~0.10% round-trip futures fee is cleared by the ~1.8-sigma reversion. Long-short; leverage stays 1.0 referenced only as a non-amplifying sizing multiplier so the unused-leverage gate does not apply. Three tunable parameters keep the optimization surface minimal; SOLUSDT.BINANCE 1H has dense multi-year history so no data_unavailable risk.

The daily-anchored-VWAP excursion mean-reversion fails catastrophically on SOL despite the redesign intended to correct the abandoned variant's failure modes. Over a dense 4,559-trade sample (5.8 years) it loses -63.2% (CAGR -16.4%) with Sharpe -1.79 and the ENTIRE CI below zero (-2.41 to -1.17), PF 0.84, probabilistic_sharpe 0.0, a 65.8% max drawdown, and losses in all 7 years. The picking-pennies death mode is fully present: win rate is high (62%) but avg_win ($118) is half avg_loss ($233) — the small VWAP reversions are dwarfed by trend-through losses, which is exactly the diagnosed failure for VWAP-MR on a HIGH-volatility asset (SOL is the worst case, excursions trend through fair value rather than revert). The long-short / no-regime-gate / modest-threshold / ATR-stop redesign did NOT fix the mechanism, and commission is still 10.5% of gross on the high-turnover trading. This is now the mechanism's THIRD documented failure across transfers: ETH long-only (abandoned), SOL up-the-curve (this run and a prior -80% variant), and BTC down-the-curve (-76.6%) — decisive evidence that daily-anchored-VWAP-MR is asset/config-specific to the original ETH strategy and does not generalize, least of all to high-vol SOL. No optimization fixes a signal losing ~16%/year with its entire Sharpe CI below zero, a 66% drawdown, and a structural picking-pennies + fee-drag loss profile. FAILURE PATTERN: intraday daily-anchored-VWAP excursion mean-reversion does not transfer off ETH; on every other major (higher-vol SOL, lower-vol BTC) the excursions trend through fair value and the high-turnover strategy bleeds via small-win/large-loss geometry plus ~10-19% fee drag, producing deeply negative Sharpes with entire CIs below zero. This mechanism should not be transferred to further assets regardless of redesign tweaks; the volatility-direction reasoning (lower-vol reverts, higher-vol trends-through) is not the lever — the edge simply does not exist outside ETH's specific regime.