XrpSpot1hDailyVwapExcursionLong

XRP BINANCE_SPOT 1H Daily-Anchored VWAP Excursion LONG-ONLY Mean-Reversion (Spot Venue Quota Fill, Single-Instrument, 3-Parameter, ATR-Relative Stop, Bounded Infra Params)

A LONG-ONLY, single-instrument intraday mean-reversion strategy on XRPUSDT.BINANCE_SPOT using 1-HOUR bars and OHLCV-only data. This deliberately targets the BINANCE_SPOT venue bucket (11.1% actual vs 15% target) — which, given this session's hard evidence, is now the ONLY under-represented venue that is reliably testable: HYPERLIQUID is data-span-blocked (ETHUSD-HL 1H had only ~7 months, SOL-HL 15m only 38 days — two consecutive HL transplant abandons), COIN-M is engine-blocked (InstrumentId '_' misclassification), options are tooling-blocked (no multi-leg/rolling-expiry booking), and cross-venue is tooling-blocked (no funding-cash-flow accounting or cross-venue MTM netting). Spot is therefore the highest-value move that both diversifies away from the 82.4% Binance-USD-M concentration AND will actually run on multi-year catalog data. It uses the ONE mechanism class that cleared every optimization gate this session — daily-anchored VWAP excursion mean-reversion (the ETH-1H Binance futures sibling was PROMOTED; range-MR, funding-MR, and the options/carry families all failed) — transplanted to (a) a venue with abundant history (XRPUSDT spot has ~6 years of continuous 1H bars, >50,000 bars, eliminating the data-insufficiency failure mode) and (b) a fresh asset outside the BTC/ETH/SOL VWAP cluster. XRP is chosen specifically because its price action is long ranging/chop periods punctuated by sharp pumps — the ideal regime for excursion MR, where below-VWAP dips revert rather than trend persistently down. It is fully testable: single-instrument LONG-ONLY directional spot PnL (CASH account, no shorting, no funding, no second leg). Two explicit lessons applied: (1) FEE VIABILITY — spot round-trip is ~0.20%, so the entry z-threshold is set HIGH (>=2.5) so the expected reversion exceeds the >0.25% spot floor with margin (no sub-0.25% scalping). (2) WIN/LOSS ASYMMETRY — the stop is ATR-relative (scales with the volatility that produced the excursion), NOT a fixed % that is tighter than the reversion target, directly fixing the inverted avg_loss/avg_win (2.05x) that sank the SOL-HL predecessor. Exactly 3 tunable parameters (z_entry, vwap_std_lookback, stop_atr_mult); all infra params (min_session_bars, time-stop bars, min_notional) are FIXED at construction and bounded to physically valid ranges — they must NOT enter the Optuna search space, closing the optimizer-escape hole where eod_exit_hour=30 voided the ETH-HL hypothesis.

Transplants the one mechanism class that cleared every optimization gate this session -- daily-anchored VWAP excursion mean-reversion (the promoted ETH-1H Binance-futures sibling) -- onto BINANCE_SPOT, the only under-represented venue that runs on abundant multi-year catalog data (XRPUSDT spot has ~6 years of continuous 1H bars), diversifying away from the Binance-USD-M concentration without the data-span / engine / tooling blocks that sank the HYPERLIQUID, COIN-M, options, and cross-venue alternatives. Venue routing is correct: long-only with no leverage and no funding edge belongs on BINANCE_SPOT (CASH account, leverage clamped to 1), so only OrderSide.BUY is ever returned and notional is capped at equity. The returned signal is the continuous VWAP z-score (varies every bar -> non-frozen on real data; gracefully 0.0/invalid before MIN_SESSION_BARS, which is non-fatal in Layer 2). Two explicit predecessor lessons are applied: (1) FEE VIABILITY -- z_entry is floored at 2.5 so reversion clears the spot fee floor with margin (no sub-0.25% scalping); (2) WIN/LOSS ASYMMETRY -- the stop is ATR-relative (stop_atr_mult x ATR captured at entry), scaling with the volatility that produced the excursion rather than a fixed % tighter than the reversion target, directly addressing the inverted avg_loss/avg_win that sank the SOL-HL predecessor. Exactly three tunable parameters (z_entry, vwap_std_lookback, stop_atr_mult); MIN_SESSION_BARS, MAX_HOLD_BARS, ATR_PERIOD and POSITION_PCT are fixed class constants bounded to valid ranges and never enter the Optuna space, closing the optimizer-escape hole (e.g. eod_exit_hour=30) that voided the ETH-HL hypothesis. The daily deviation deque is cleared at each UTC-midnight re-anchor so the std never mixes two VWAP anchors, and the z is marked invalid until MIN_SESSION_BARS samples exist to avoid a spurious z=0 'revert' exit right after the reset.

Decisively negative, fee-fragile edge on a large, well-covered sample — a structural failure, not a data problem. Across 1911 trades over ~7.5 years (2018-2026, good coverage): Sharpe -2.49, Sortino -2.37, PSR 0.0 with the entire Sharpe CI negative (ci_high -1.79), profit_factor 0.67, expectancy -$28/trade, total_return -41%, max DD 41%, and NEGATIVE in all 8 years. Two structural problems optimization cannot fix: (1) negative payoff asymmetry — win rate 59% but avg_loss $212 is 2.1x avg_win $99 (return_skew -2.62); the ATR-relative stop did NOT solve the inverted ratio because long-only 'buy the dip below VWAP' on XRP gets steamrolled when XRP dumps (the dip keeps falling to the ATR stop), so small capped reversion wins lose to larger trend-continuation stops. (2) Fee-fragility — commission_pct_of_gross 25.8%, so even the high z=2.5 gate's reduced trade count cannot escape the ~0.20% spot round-trip. The transplant is also flawed: the promoted ETH-1H sibling was a FUTURES long+short, and dropping to long-only spot removes the short side that captured half the excursions while adding the spot fee drag. Unlike the SOL-15m / ETH-HL data-iterate cases, the data coverage here is excellent, so this is purely a negative-edge abandon; the code faithfully implements the premise, which simply loses on XRP spot. FAILURE PATTERN: long-only intraday daily-anchored VWAP-excursion mean-reversion on a single high-volatility alt (XRP spot 1H) has a negative payoff asymmetry (small capped reversion wins vs larger trend-continuation stops when the asset dumps) that, combined with ~0.20% spot fees eating ~26% of gross, loses in every year; a higher z-threshold and ATR stop do not rescue it, and dropping the short side of the promoted futures sibling removes half the edge.

Long-only intraday mean-reversion on XRPUSDT.BINANCE_SPOT 1-hour bars using a daily-anchored VWAP. Each UTC day it anchors a volume-weighted average price from 00:00 UTC and maintains a rolling window (vwap_std_lookback bars, reset at each daily re-anchor) of the deviation (close - VWAP), forming z = (close - VWAP)/std. It buys when z <= -z_entry (price at least z_entry sigma below the day VWAP) and the 14-bar ATR is positive. It exits on the earliest of: z returning to >= 0 (price back at/above VWAP), an ATR-relative stop (close <= entry_price - stop_atr_mult x ATR-at-entry), or a 24-bar time stop. Sizing is a fixed 10% of equity, long-only, CASH/1x (spot cannot short or use leverage). The z-entry is held at >= 2.5 so the expected reversion exceeds the ~0.25% spot round-trip fee floor.

Excellent data coverage (1911 trades over ~7.5 years, >69k bars) — the hypothesis succeeded at avoiding the HL data-insufficiency, making the result conclusive

Disciplined 3-tunable design with fixed infra params and a high z-gate; honest metrics (metrics_reliable true)

Decisively negative: Sharpe -2.49, Sortino -2.37, PSR 0.0, entire Sharpe CI negative (ci_high -1.79), total_return -41%, NEGATIVE in all 8 years

profit_factor 0.67, expectancy -$28/trade; negative payoff asymmetry — win rate 59% but avg_loss $212 is 2.1x avg_win $99 (return_skew -2.62), the ATR stop did not fix it

Fee-fragile: commission_pct_of_gross 25.8% — ~0.20% spot RT eats a quarter of gross even at the high z=2.5 gate

Transplant error: the promoted ETH-1H sibling was FUTURES long+short; long-only spot drops the short side and adds fee drag, and XRP's hard dumps steamroll the buy-the-dip leg

XrpSpot1hDailyVwapExcursionLong transplanted the promoted ETH-1H VWAP-excursion mechanism onto XRP spot — the one under-represented venue with abundant multi-year data after Hyperliquid, COIN-M, options, and cross-venue all proved blocked — with a high z-gate and ATR stop meant to fix prior fee and asymmetry failures. The data coverage was excellent (1,911 trades over 7.5 years), making the verdict conclusive: it lost -41% with a -2.49 Sharpe and 0.67 profit factor, negative in every year, because long-only buy-the-dip on XRP gets run over when it dumps and spot fees consumed a quarter of gross. The analyst flagged it as a structural negative-edge transplant error — dropping the short side of a long+short futures sibling removed half the edge — and abandoned it before optimization. It ended after one iteration as abandoned, never advancing to optimization or risk review.

Long-only intraday VWAP-excursion mean-reversion on a single high-vol alt spot has a negative payoff asymmetry (small capped reversion wins vs larger trend-continuation stops when the asset dumps) that a higher z-threshold and ATR stop cannot rescue, and ~0.20% spot fees eat ~26% of gross — and dropping the short side of a promoted long+short futures sibling removes half its edge, so a venue/direction transplant of a proven mechanism is not automatically viable.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) as a structural negative-edge failure on excellent data — the ATR stop did not fix the asymmetry because long-only buy-the-dip gets steamrolled when XRP dumps (the dip falls to the stop), spot fees ate a quarter of gross, and dropping the short side of the futures sibling removed half the edge — so optimization and all later stages were never reached.

A long-only intraday daily-anchored-VWAP excursion mean-reversion on XRPUSDT.BINANCE_SPOT (1H bars, OHLCV-only, CASH 1x) that bought when price extended >=2.5 sigma below the session VWAP and exited on reversion to VWAP, an ATR-relative stop, or a 24-bar time cap — transplanting the promoted ETH-1H VWAP-excursion mechanism onto the only reliably-testable under-represented venue (spot), with a high z-gate for fee viability and an ATR stop to fix a prior win/loss asymmetry.

Over ~7.5 years and a large 1,911-trade sample it was decisively negative: total return -41%, Sharpe -2.49 (entire CI negative, ci_high -1.79), Sortino -2.37, PSR 0.0, profit factor 0.67, expectancy -$28/trade, max drawdown 41%, and negative in all 8 years — with a negative payoff asymmetry (59% win rate but avg loss $212 = 2.1x avg win $99, return_skew -2.62) and commissions consuming 25.8% of gross.