Get Basic Confluence Matrix

6×6 basic-signal agreement matrix — pairwise directional confluence.

Parallel to /api/signals/advanced/confluence-matrix, scoped to the six Basic Signals persisted by BasicSignalEngine. Answers: which of my basic signals normally agree, and where is an unusual divergence right now? Continuous directional reads (no triggered events) — every non-zero snapshot contributes to agreement/disagreement counts.

Logic (src/api/database.py): Joins signal_scores × signal_component_scores for the last lookback timestamps, filtered to the six basic signal names. Each score is bucketed +1 / 0 / −1 with neutral_epsilon = 0.02. For each ordered pair (c1, c2):

• observations = cycles where both signals have data.
• active_observations = cycles where both signs are non-zero.
• agreement = same non-zero sign; disagreement = opposite non-zero signs.

Params:

• symbol (default SPY).
• lookback — 10–2000, default 120. Number of recent timestamps.

Signals (6, fixed order): tape_flow_bias, skew_delta, vanna_charm_flow, dealer_delta_pressure, gex_gradient, positioning_trap.

Returns:

{
  "underlying": "SPY",
  "lookback": 120,
  "components": ["tape_flow_bias", "skew_delta", "vanna_charm_flow",
                 "dealer_delta_pressure", "gex_gradient", "positioning_trap"],
  "row_order":  ["tape_flow_bias", "skew_delta", "...4 more..."],
  "matrix": {
    "tape_flow_bias": {
      "skew_delta": {
        "observations": 118,
        "active_observations": 92,
        "agreement_count": 74,
        "disagreement_count": 18,
        "neutral_count": 26,
        "agreement_ratio": 0.8043,
        "disagreement_ratio": 0.1957,
        "net_confluence": 0.6087
      }
    }
  },
  "sample_count": 118,
  "latest_timestamp": "2026-04-22T18:55:00Z"
}

• components / row_order — length-6 string array; axis labels.
• matrix[c1][c2].observations — int, [0, lookback]; cycles where both had data.
• matrix[c1][c2].active_observations — int, [0, observations]; both non-zero.
• matrix[c1][c2].agreement_count — int, [0, active_observations].
• matrix[c1][c2].disagreement_count — int, [0, active_observations].
• matrix[c1][c2].neutral_count — int, [0, observations]; at least one side was 0.
• matrix[c1][c2].agreement_ratio — [0, 1] | null if active == 0; 4 decimals.
• matrix[c1][c2].disagreement_ratio — [0, 1] | null.
• matrix[c1][c2].net_confluence — [-1, +1]; 0.0 if active == 0. Heatmap value.
• sample_count — int; distinct timestamps aggregated.
• latest_timestamp — ISO-8601 UTC | null if empty.

Trader interpretation.

• net_confluence > +0.5 — pair agrees > 75% of active cycles. Live disagreement from such a pair is a flag.
• net_confluence < −0.3 — pair structurally disagrees. Live agreement = unusual consensus.
• active_observations < ~20 — weak sample; discount.
• Diagonal cells always net_confluence = 1.0; sanity check only.

Page design. 6×6 heatmap with diverging color scale (green = +1 agree, white = 0, red = −1 disagree). Cell tooltip shows agreement_ratio / disagreement_ratio / observations. Sort rows by average agreement_ratio to surface consensus signals at top, outliers at bottom. Symmetric — render only the upper triangle if space-constrained.