PREDICTAX9 ⚽🔮

World Cup AI Prediction Engine Built for the Stair AI Agent Arena

---

What Is This?

PredictaX9 is an AI-powered football prediction engine built specifically for the 2026 FIFA World Cup. It predicts match outcomes and goal markets with mathematical consistency — every probability it outputs is logically sound and verifiable.

Built as a participant agent for the Stair AI World Cup Agent Arena, where AI agents compete by making real prediction market bets on World Cup matches, with their reasoning tracked and audited live.

---

How It Works

Most football prediction models train separate classifiers for every market independently. This causes a fundamental problem — the probabilities contradict each other and can sum to over 100%, which is mathematically impossible.

SupaTX Oracle solves this with a two-layer architecture:

┌─────────────────────────────────────────────────────────┐
│                   INPUT LAYER                           │
│   Team form, H2H history, venue context                 │
└────────────────────┬────────────────────────────────────┘
                     │
          ┌──────────┴──────────┐
          │                     │
          ▼                     ▼
┌─────────────────┐   ┌─────────────────────┐
│  POISSON MODEL  │   │  MULTICLASS MODEL   │
│                 │   │                     │
│  Predicts:      │   │  Predicts:          │
│  λ home goals   │   │  Home Win / Draw /  │
│  λ away goals   │   │  Away Win           │
│                 │   │  (softmax — always  │
│                 │   │   sums to 100%)     │
└────────┬────────┘   └──────────┬──────────┘
         │                       │
         ▼                       ▼
┌─────────────────┐   ┌─────────────────────┐
│  PURE MATH      │   │  DERIVED MARKETS    │
│                 │   │                     │
│  Over/Under 0.5 │   │  1X (Home or Draw)  │
│  Over/Under 1.5 │   │  X2 (Draw or Away)  │
│  Over/Under 2.5 │   │  12 (Home or Away)  │
│  Over/Under 3.5 │   │                     │
│  Over/Under 4.5 │   │  Always consistent  │
│  BTTS           │   │  Always sum to 100% │
│  Clean Sheets   │   │                     │
└─────────────────┘   └─────────────────────┘

Why This Matters

• Over 2.5 + Under 2.5 = exactly 100%. Always.
• Home Win + Draw + Away Win = exactly 100%. Always.
• No contradictions. No impossible probabilities. Just clean math.

---

Sample Output

=========================================================
  ⚽ PredictaX9: Brazil vs Argentina
=========================================================

🔵 Expected Goals
   Home : 1.515
   Away : 1.107
   Total: 2.62

🟡 Over/Under  (each row sums to 100%)
   Threshold    Over      Under      Sum
   ────────────────────────────────────
   0.5        92.7%      7.3%    100.0% ⭐
   1.5        73.7%     26.3%    100.0% ⭐
   2.5        48.7%     51.3%    100.0%
   3.5        26.9%     73.1%    100.0% ⭐
   4.5        12.6%     87.4%    100.0% ⭐

🟢 Match Outcome  (sums to 100%)
   Home Win  :  44.3%
   Draw      :  21.3%
   Away Win  :  34.3%
   ─────────────────────────
   1X (H/D)  :  65.7%
   X2 (D/A)  :  55.7%
   12 (H/A)  :  78.7%

🔴 Other Markets
   BTTS            :  52.2%
   Home Clean Sheet:  33.1%
   Away Clean Sheet:  22.0%

⭐ HIGH CONFIDENCE PICKS  (≥65.0%)
   ✅ Over_0.5           →  92.7%  ██████████████████
   ✅ Under_4.5          →  87.4%  █████████████████
   ✅ 12                 →  78.7%  ███████████████
   ✅ Over_1.5           →  73.7%  ██████████████
   ✅ Under_3.5          →  73.1%  ██████████████
   ✅ 1X                 →  65.7%  █████████████
=========================================================

---

Model Performance

Trained on 49,000+ international football matches from 1872 to 2026, filtered for competitive fixtures (World Cup, qualifiers, continental championships).

Model	Metric	Score
Home Goals (Poisson)	MAE	~0.55
Away Goals (Poisson)	MAE	~0.55
Match Outcome (Multiclass)	Accuracy	~52%
Match Outcome (Multiclass)	Log Loss	~1.00

Match outcome accuracy of ~52% significantly beats the random baseline of 33% for a 3-class problem (Home/Draw/Away).

---

Project Structure

predictax9/
├── predictax9.py     ← Main prediction engine (run this)
├── README.md               ← You are here
├── requirements.txt        ← Dependencies
└── predictax9_models/          ← Trained model files (not included)
    ├── feature_cols.pkl
    ├── model_home_goals.pkl
    ├── model_away_goals.pkl
    └── model_outcome.pkl

Note: The supatx_models/ folder is not included in this repository. You must train the models yourself using the training pipeline. This protects the integrity of the trained weights.

---

Installation

# Clone the repo
git clone https://github.com/Supatx/predictaX9.git
cd predictax9

# Install dependencies
pip install -r requirements.txt

---

Usage

1. Train the models using the training pipeline
2. Place the predictaX9_models/ folder in the project root
3. Open predictaX9.py and fill in match stats at the bottom
4. Run:

python predictaX9.py

---

Required Input Data

To predict a match you need these 10 values. All available from FIFA.com, Sofascore, or FBREF:

Input	Description	Example
home_goals_scored_last5	Avg goals HOME scored in last 5 games	2.2
home_goals_conceded_last5	Avg goals HOME conceded in last 5 games	0.8
home_win_rate_last5	HOME win rate last 5 games (0.0–1.0)	0.8
away_goals_scored_last5	Avg goals AWAY scored in last 5 games	1.4
away_goals_conceded_last5	Avg goals AWAY conceded in last 5 games	1.2
away_win_rate_last5	AWAY win rate last 5 games (0.0–1.0)	0.6
h2h_avg_goals	Avg total goals in last 5 H2H meetings	2.5
h2h_avg_btts	Fraction of H2H where both teams scored	0.7
neutral_venue	Neutral ground? World Cup = always 1	1
went_to_shootout	Upcoming match = always 0	0

---

Requirements

lightgbm
pandas
numpy
scipy
joblib
scikit-learn

---

Built For

This agent is being entered into the Stair AI World Cup Agent Arena — a live competition during the 2026 FIFA World Cup where AI agents make real prediction market bets and are scored on both profit and reasoning quality.

---

Author

Elijah Olalere — @SupaTX

Building AI agents that think clearly and predict intelligently.

---

License

MIT License — free to use, learn from, and build on.