World Cup 2026 — Prediction System

AI-powered match analysis using the Game Sense framework.

---

HOW TO LOAD CONTEXT (keep it lean)

For any match analysis, load ONLY these files:

1. _skill.md — the prediction framework (phases 1-6)
2. _groups.md — current group standings (for motivation mapping)
3. teams/XXX.md — home team profile
4. teams/YYY.md — away team profile

Do NOT load the full teams folder. Load only the two relevant team files.

---

POST-MATCH WORKFLOW (run after every result)

1. Open the two team files → update Tournament Form section with result + xG + key obs
2. Open _model-log.md → add row to backtesting table
3. Open _groups.md → update the affected group table
4. Open web/data.js → update match entry from pending to completed with result
5. (Optional) Save pre-match analysis to analysis/YYYY-MM-DD_XXX-YYY.md

Do not update the HTML file. Only data.js changes. The page re-renders automatically.

---

TEAM FILE NAMING

Use 3-letter FIFA codes: ARG, FRA, EGY, NZL, URU, CPV, BRA, GER, etc.

---

DEPLOYING THE WEB PAGE

The web/ folder is a self-contained deployable unit:

• Upload web/index.html + web/data.js to any static host
• GitHub Pages, Netlify, Vercel — all work with zero config
• After each game: update data.js only, re-upload that one file

---

GROUPS QUICK REF

Group	Teams
G	Egypt, Belgium, Iran, New Zealand
H	Spain, Uruguay, Cape Verde, Saudi Arabia
I	France, Norway, Senegal, Iraq
J	Argentina, Austria, Algeria, Jordan
K	Colombia, Portugal, Congo DR, Uzbekistan
L	England, Panama, Ghana, Croatia

---

MODEL PERFORMANCE SUMMARY

→ See _model-log.md for full breakdown.