UK49sStatistical Research & Audit

System

Settings

Connection, appearance and the methodological definitions that govern every figure displayed in this application.

API connection

Configured via the VITE_API_URL environment variable

Base URL

https://uk49s.drapeasy.co.za/api

Connectivity

checking…

Feed status

Last successful import

Live endpoints: GET /api/dashboard, GET /api/draws/:type, GET /api/draws/:type/latest, POST /api/admin/refresh. Not yet implemented server-side: /api/stats/numbers, /api/models, /api/predictions, /api/backtest, /api/quality — these render labelled placeholder states rather than substituted values, with the ledger and backtesting views reading the predictions and performanceSummary arrays from /api/dashboard instead.

Baselines and thresholds

Single-number baseline

12.2%

6/49 per draw

Top-3 baseline

33.2%

≥1 of 3 candidates drawn

Minimum sample size

30

below this: insufficient data

Status badge semantics

  • Insufficient data n below 30: no comparison is drawn.
  • Below baseline Measured rate materially under the baseline.
  • At baseline Indistinguishable from the baseline.
  • Above baseline — unvalidated Above baseline in testing only — never green.
  • Validated above baseline Reserved for confirmed out-of-sample results with a significant p-value.

Statistical observation

Descriptive summary of historical records. Does not predict future outcomes.

Experimental model — unvalidated

In testing. No result here is confirmed out-of-sample.

Out-of-sample audit

Pre-draw records only, scored after the draw. No post-draw fitting.

Terminology policy

Not used anywhere in this platform

  • outcome-claim phrasing about draw results
  • guarantees or certainty language
  • "banker" / confidence-pick phrasing
  • accuracy claims about future draws

Used instead

  • statistical ranking
  • top-ranked candidate
  • historical frequency
  • model output
  • audit trail
theme: dark default, light optionalnumbers set in monospaceno gambling promotion

This platform is a research and audit tool. It reports what has been recorded and how model output performed under audit — nothing more.