{
  "version": "1.0",
  "strategy": "vwap_pull_in",
  "title": "RVOL x Instrument-Type Cross-Tab — does RVOL add edge after controlling for type?",
  "generated": "2026-06-09",
  "analysis_type": "read_only_no_db_writes_validation_addon",
  "core_question": "Does RVOL add independent signal once instrument type is controlled?",
  "answer": "NO. RVOL does not survive instrument-type control. Within megacap-tech the <1.0 bucket is the BEST (+$145.86/62.5%); the headline '<1.0 is worst' was an IBIT/crypto-proxy confound (IBIT alone = -$347 over 4 low-RVOL trades, incl. the -$216 outlier). Fade loses and expansion wins at every RVOL bucket. RVOL fails even as a floor.",

  "data_basis": {
    "source": "Z:/trading/dev/vwap-band-entry-surface (Liberator backfill; same artifacts as the gate validation audit)",
    "files": ["stage_b/trades_full.csv", "stage_b/ENTRY_LOCATION_FEATURES_FULL_WINDOW.csv (first_5_min_state join)"],
    "no_db_query": true,
    "trade_set": "plain vwap-pull-in, n=132 (net -$298.13, full Liberator window)"
  },

  "coverage": {
    "total_plain_pi": 132,
    "rvol_populated": 108,
    "rvol_populated_pct": 81.8,
    "gap_populated": 132,
    "instrument_type_classified_non_other": 107,
    "other_unknown_n": 25,
    "instrument_type_map_is_heuristic": true
  },

  "instrument_type_overall_plainPI": {
    "megacap_tech": {"n": 51, "win_rate": 51.0, "total_pnl": 151.91, "avg": 2.98, "tier": "only net-positive type"},
    "crypto_proxy_etf": {"n": 31, "win_rate": 38.7, "total_pnl": -179.02, "avg": -5.77},
    "smallcap_highbeta": {"n": 16, "win_rate": 31.2, "total_pnl": -113.73, "avg": -7.11},
    "energy_commodity": {"n": 9, "win_rate": 11.1, "total_pnl": -57.07, "avg": -6.34, "tier": "insufficient_sample"},
    "other_unknown": {"n": 25, "win_rate": 12.0, "total_pnl": -100.22, "avg": -4.01}
  },

  "Q1_within_type_does_rvol_separate": {
    "verdict": "NO - RVOL is non-monotonic / noise within type",
    "megacap_tech_type_x_rvol": {
      "lt1.0": {"n":16,"win_rate":62.5,"total_pnl":145.86},
      "1.0_1.5": {"n":13,"win_rate":38.5,"total_pnl":-63.75},
      "1.5_2.5": {"n":9,"win_rate":55.6,"total_pnl":73.57},
      "gt2.5": {"n":5,"win_rate":40.0,"total_pnl":-18.15}
    },
    "note": "Within the GOOD type, the gate's ordering inverts: <1.0 and 1.5-2.5 win, 1.0-1.5 (the 'green' band) loses. No usable RVOL ordering."
  },

  "Q2_does_lt1_stay_bad_after_type_control": {
    "verdict": "NO - the <1.0 badness is a type+symbol confound",
    "floor_test_lt1_vs_ge1": {
      "megacap_tech": {"lt1.0": {"n":16,"win_rate":62.5,"total_pnl":145.86}, "ge1.0": {"n":27,"win_rate":44.4,"total_pnl":-8.33}},
      "crypto_proxy_etf": {"lt1.0": {"n":12,"win_rate":41.7,"total_pnl":-292.85}, "ge1.0": {"n":13,"win_rate":38.5,"total_pnl":7.67}},
      "smallcap_highbeta": {"lt1.0": {"n":8,"win_rate":12.5,"total_pnl":-84.87}, "ge1.0": {"n":2,"win_rate":50.0,"total_pnl":-19.95}},
      "energy_commodity": {"lt1.0": {"n":7,"win_rate":14.3,"total_pnl":-43.25}, "ge1.0": {"n":1,"total_pnl":-2.5}},
      "other_unknown": {"lt1.0": {"n":10,"win_rate":0.0,"total_pnl":-45.51}, "ge1.0": {"n":12,"win_rate":16.7,"total_pnl":-44.62}}
    },
    "smoking_gun": "In megacap-tech, <1.0 is +$145.86 (BEST). The entire negative <1.0 sits in crypto-proxy (-$292.85) and other.",
    "lt1.0_loss_drivers": {"IBIT": {"type":"crypto_proxy_etf","n":4,"pnl":-347.0,"note":"includes the -$216 6/3 outlier"}, "RGTI":{"n":3,"pnl":-38.98}, "CVX":{"n":4,"pnl":-33.31}, "KOPN":{"n":1,"pnl":-28.8}}
  },

  "Q3_does_high_rvol_underperform_within_type": {
    "verdict": "NO consistent high-RVOL penalty; old story was outlier/symbol-driven",
    "evidence": "megacap >2.5 n=5 -$18; crypto >2.5 n=2 +$15; smallcap >2.5 n=1 -$28; other >2.5 n=5 -$27. All cells tiny. On ALL trades the gt2.5 x gap>1.5 cell is +$223.83/n27 (high RVOL is NOT intrinsically bad).",
    "tier": "insufficient_sample_per_cell_but_no_signal"
  },

  "Q4_is_rvol_useful_as_a_floor": {
    "verdict": "NO - the floor is also confounded and would harm the good type",
    "reason": "A <1.0 cutoff would wrongly EXCLUDE megacap-tech's best trades (+$145.86/62.5%). The floor only 'works' by proxy-excluding crypto-proxy/IBIT, which instrument-type + symbol rules already handle better and more honestly.",
    "tier": "rejected"
  },

  "Q5_rvol_x_gap_grid_plainPI": {
    "verdict": "Gap separates within EVERY RVOL row; RVOL does not separate within gap columns (except the IBIT-driven lt1.0 x le0.5 cell).",
    "gap_gt1.5_is_negative_at_every_rvol": {"lt1.0":-121.27,"1.0_1.5":-60.75,"1.5_2.5":-84.65,"gt2.5":-43.41,"missing":-31.85},
    "lt1.0_x_le0.5_cell": {"n":17,"win_rate":41.2,"total_pnl":-229.31,"note":"the 'calm+calm disaster' cell = IBIT/crypto confound, not an RVOL effect"}
  },

  "Q6_rvol_x_post_entry_state_plainPI": {
    "verdict": "Post-entry state DOMINATES RVOL. Fade loses and expansion wins at every RVOL bucket.",
    "fade_by_rvol": {"lt1.0":{"n":18,"win_rate":11.1,"total_pnl":-400.08},"1.0_1.5":{"n":8,"win_rate":0.0,"total_pnl":-103.52},"1.5_2.5":{"n":6,"win_rate":16.7,"total_pnl":-56.12},"gt2.5":{"n":3,"win_rate":0.0,"total_pnl":-36.98}},
    "early_expansion_by_rvol": {"lt1.0":{"n":5,"win_rate":80.0,"total_pnl":55.64},"1.0_1.5":{"n":3,"win_rate":66.7,"total_pnl":53.98},"1.5_2.5":{"n":5,"win_rate":80.0,"total_pnl":4.9},"gt2.5":{"n":3,"win_rate":66.7,"total_pnl":19.98}},
    "note": "RVOL adds nothing on top of the post-entry state signal."
  },

  "does_rvol_survive": false,
  "rvol_survives_after_type_control": "NO",

  "strongest_positive_cells": [
    {"cell": "megacap_tech x RVOL<1.0", "n": 16, "win_rate": 62.5, "total_pnl": 145.86, "meaning": "low RVOL is fine when the name is megacap-tech - kills the 'calm=bad' story"},
    {"cell": "plain-PI early_expansion (any RVOL)", "win_rate_range": "66-80%", "meaning": "post-entry expansion wins regardless of RVOL"},
    {"cell": "megacap_tech overall", "n": 51, "win_rate": 51.0, "total_pnl": 151.91, "meaning": "the only net-positive instrument type"}
  ],
  "strongest_negative_cells": [
    {"cell": "crypto_proxy_etf x RVOL<1.0", "n": 12, "total_pnl": -292.85, "meaning": "the <1.0 'disaster' is here, driven by IBIT (-$347/4 incl -$216 outlier)"},
    {"cell": "plain-PI fade (any RVOL)", "total_pnl": "-$36 to -$400", "meaning": "fade loses at every RVOL bucket"},
    {"cell": "any RVOL x |gap|>1.5%", "meaning": "negative at every RVOL bucket - gap is the real RED"}
  ],

  "recommendation_choice": "D",
  "recommendation_detail": "D - Remove RVOL from the Green/Yellow/Red gate until a cleaner, type-controlled RVOL backfill demonstrates it adds something. RVOL does not survive type control and fails even as a floor. RVOL MAY be retained ONLY as a non-gating liquidity/activity context readout (the legitimate kernel of option C) - never as a winner/loser threshold. Reject the '<1.0 = avoid' rule (confounded). This matches the priority order: gap first, instrument type second, post-entry behavior third, RVOL only as activity/liquidity context.",
  "options_considered": {
    "A_keep_rvol_major_axis": "REJECTED - no independent signal",
    "B_demote_to_secondary_confidence": "partial - acceptable but still implies edge value RVOL doesn't have",
    "C_floor_only": "REJECTED as an edge floor - the <1.0 floor is confounded and harms megacap-tech; only the pure-liquidity (tradability) reading survives",
    "D_remove_from_gate_keep_as_context": "CHOSEN"
  },

  "revised_quality_gate": {
    "priority_order": ["gap", "instrument_type", "post_entry_behavior", "rvol_context_only"],
    "RED_primary": "abs(gap_pct) > 1.5%",
    "RED_secondary": "instrument type in {crypto_proxy_etf, smallcap_highbeta, energy_commodity, other_unknown} UNLESS post-entry expansion confirms (megacap_tech is the only net-positive type in-sample)",
    "GREEN": "controlled gap (abs<=1.5%, ideally <=0.5%) AND megacap-tech AND no 3-min fade",
    "YELLOW": "controlled gap but lower-confidence instrument type OR missing data",
    "HARD_EXIT_RED_FLAG": "3-minute fade (wins ~7-11% at every RVOL bucket)",
    "TREND_HOLD_ONLY_IF": "10-15 min expansion confirmation + slope acceleration + CBO/range-break evidence if available",
    "RVOL_role": "non-gating liquidity/activity context display only; NOT a go/no-go input"
  },

  "validation_status": {
    "rvol_as_gate_axis": "rejected",
    "rvol_lt1_avoid_floor": "rejected",
    "rvol_as_liquidity_context_only": "working_rule_needs_backtest",
    "gap_gt1.5_RED": "promoted_candidate",
    "instrument_type_axis": "promoted_candidate (megacap-tech eligible; rest stricter)",
    "3min_fade_exit": "promoted_candidate",
    "10_15min_expansion_trend_hold": "promoted_candidate",
    "energy_commodity_and_other_cells": "insufficient_sample",
    "overall": "in-sample only (~2mo, one trader); needs out-of-sample/walk-forward before promotion"
  },

  "sample_size_warnings": [
    "Per-(type x RVOL) cells are mostly n<10; energy_commodity (n=9 total) and smallcap (n=16) are thin; gt2.5 cells n=1-5.",
    "RVOL populated 108/132 (82%); 24 NA. other_unknown = 25/132 (19%) unclassified.",
    "The decisive megacap <1.0 = +$145.86 cell is n=16 (modest but interpretable); the IBIT -$347 driver is concrete (n=4, one -$216 outlier).",
    "Instrument-type map is heuristic; TSLA classed megacap_tech, MRVL/SNOW/CRWV classed smallcap_highbeta - reasonable people could reclassify. Full map embedded for transparency.",
    "No out-of-sample / walk-forward. Treat as hypotheses (CMT Ch.35/38 data-mining caution)."
  ],

  "production_untouched": ["/strategies/", "/scorecards/", "gen_scorecard.py", "_scientific_overlay.js", "taxonomy", "DB", "no Stage B reclassification"]
}
