Problem Formulation & Targets¶

Objective¶

The goal of the system is to predict the outcome of football matches based on historical and contextual match data.

ML task definition¶

Depending on the experiment, the task is formulated as:

• Classification Predict discrete match outcomes (e.g., home win / draw / away win).

• Regression Predict continuous signals (e.g., expected goals, score difference).

Target construction¶

Targets are constructed using: - final match results, - clearly defined temporal cutoffs.

All targets satisfy: - no future information leakage, - deterministic construction from raw data.

Business constraints (translated into ML)¶

• predictions must be available before match start,
• inference latency must be compatible with interactive use,
• models must generalize across seasons and competitions.

Out of scope¶

• live in-play prediction,
• betting strategy optimization,
• financial outcome modeling.