Research Analytics Engine · v24.4 · Read-only intelligence over prediction history
Research Lab is read-only. It does not modify V1, V2, slips, confidence scores, or the live engine. All analytics are computed at query time from SQLite predictions (read-only).
Note: Each prediction record covers one market for one fixture. A single fixture generates up to 15 prediction records (one per market). Settled count = prediction records settled, not fixture count.
A — Market Performance
— settled predictions across 0 markets
Loading…
Market
Samples
Won
Lost
Void
Avg Conf
Avg Odds
Win Rate
B — Confidence Calibration
Are confidence scores accurate? Positive diff = model under-confident. Negative diff = over-confident.
Loading…
Band
Samples
Won
Lost
Expected
Actual
Diff
Expected = midpoint of confidence band. Diff = actual − expected.
C — League Performance
— leagues with settled data
Loading…
League
Country
Samples
Won
Lost
Avg Conf
Win Rate
D — V2 Analytics
— V2-classified settled predictions
Loading…
Classification
Samples
Won
Lost
Avg Conf
Win Rate
E — FBIL Behaviour Tag Analytics
— FBIL-profiled settled predictions · Baseline: —
Loading…
Behaviour Tag
Samples
Won
Lost
Avg Conf
Win Rate
Tick on bar = overall baseline (—%). Tag present on either home or away team counts for that prediction.
F — SIL Standings Incentive Analytics
— SIL-profiled settled predictions · Baseline: —
Loading…
Incentive Level
Samples
Won
Lost
Avg Conf
Win Rate
G — Trend Analytics
Win rate trends over time
Loading…
V2 VALIDATION EXPANSION — v24.0
Read-only validation over existing V2 shadow classifications. Evidence over commentary.
H — Market Fit Validation (Global)
Baseline: — · — V2 settled
Loading…
No data yet.
Descriptive analytics only. Not promotion. Does not modify predictions.
H₂ — Market Fit by Market
Does Market Fit create value equally across markets?
Loading…
Market
Tier
Samples
Win Rate
Avg Conf
Baseline Diff
Calibration
Descriptive analytics only. Not promotion. Does not modify predictions.
I — Result Character Validation
Loading…
No data yet.
Descriptive analytics only. Not promotion. Does not modify predictions.
J — Risk Framework Validation
Loading…
No data yet.
Descriptive analytics only. Not promotion. Does not modify predictions.
K — Archetype Cross Analysis
Loading…
No cross data yet.
Descriptive analytics only. Not promotion. Does not modify predictions.
L — FBIL Tag & Pair Analytics
— FBIL samples
Loading…
No FBIL data yet.
Descriptive analytics only. Not promotion. Does not modify predictions.
M — SIL Incentive Cross Analytics
— SIL samples
Loading…
No SIL data yet.
Descriptive analytics only. Not promotion. Does not modify predictions.
N — League-Level V2 Validation
Filterable league slices
Loading…
Select a league to view V2 dimension breakdowns.
Descriptive analytics only. Not promotion. Does not modify predictions.
O — Prediction Failure Analytics
— V2 losses analyzed
Loading…
Classification
Loss Count
% of Losses
Avg Conf
Overconfidence
Descriptive analytics only. Not promotion. Does not modify predictions.
Descriptive analytics only. Not promotion. Does not modify predictions.
Q — Evidence Summary
Facts only — no narrative conclusions
Loading…
—
N/A
—
N/A
—
N/A
—
N/A
—
N/A
—
N/A
—
N/A
Descriptive analytics only. Not promotion. Does not modify predictions.
R — Promotion Framework
Shadow-only infrastructure · V2 influence on V1: NONE
Loading…
Foundation release only. No V1 scoring, confidence, candidate selection, slips, or settlement changes. Shadow adjustments are stored for audit — never applied.
S — 1UP Analytics
Home/Away 1UP — wins if selected team ever held a lead (v24.4 research markets)
T — Combo Market Analytics
Double Chance + Goals combos — generated when both components exist (v24.4)