WRI9A.MAWRI9A.MAINTELLIGENCE • CULTURE • DATA
Methodology

How WRI9A generates analysis and signals

This page explains the desk in ordinary language. Live matches change; the method stays. If the API is quiet today, you can still understand what WRI9A would publish tomorrow.

What makes WRI9A different

Many prediction sites only list a pick. WRI9A writes a briefing around form, home and away performance, head-to-head notes when stored, expected-goal or Elo context, and a sentence of model reasoning. Settled signals stay on the Track Record, wins and losses together. News is summarised and linked, not copied in full.

Data sources

When a source has too little history (new team, sparse Botola sample, missing odds), the briefing says so and the model may use a conservative fallback. We would rather publish a cautious estimate than invent certainty.

Match analysis

For competitions WRI9A follows closely, a briefing typically includes:

That analysis is original WRI9A desk copy generated from our stored numbers. It is not a paste of a third-party preview. Open the live board on Football Analysis.

Model signals

Botola and Saudi Pro League matches use WRI9A ratings and a Poisson scoring model. Other competitions may use implied probabilities from bookmaker prices, or a conservative fixture fallback when prices are missing. Each signal stores selection, market, probability, risk band and model version before kick-off.

After full time, the worker settles Over/Under and “to win” selections against the official score. Pending rows are unsettled. Settled rows are won or lost. Nothing is deleted because it lost.

How to interpret a percentage

A model probability is a ranking tool. It is not a promise. Combine it with the briefing, then with your own watching. Read Responsible Betting before you attach money to any idea.

Update cadence

A scheduled job runs about every 20 minutes. Pages show a last-updated timestamp from the database when data exists. We do not display a fake clock.