Author: Mauro Mandorino et al.
Journal: Applied Sciences – 2026 (16)

The Problem: The “60-Minute” Guesswork

Coaches often ask, “How many minutes does he have in him?” especially when a player is returning from injury or has had a congested fixture schedule. Usually, this is decided by “gut feeling.” This study creates a mathematical framework to turn GPS and heart rate data into a specific Minute-Based Fatigue Threshold.

The “Fitness vs. Fatigue” Balance

The study uses a “Banister-like” model, which views a player’s readiness as a seesaw:

• Fitness (The Gain): Built up over weeks of consistent training.
• Fatigue (The Drain): The immediate “cost” of recent high-intensity sessions.
• The Goal: To identify the exact moment during a match where a player’s accumulated fatigue outweighs their current fitness, leading to a “performance drop-off.”

The model works through the following practical steps:

1. Data Collection and Processing

The system integrates several data streams from elite football players using microtechnology like GPS and heart rate (HR) monitors:

• External Load: Metrics such as total distance, high-speed running, and “mechanical work” (accelerations and decelerations) are tracked over rolling 7-day (acute) and 28-day (chronic) windows.
• Internal Load: Heart rate is collected to assess the physiological cost of the work performed.

2. Deriving Fitness and Fatigue Indices

Using machine learning (specifically Random Forest models), the system calculates two key individualized indices:

• Fitness Index (z-FI): This reflects whether a player’s heart rate is lower than expected for a given workload. A higher score suggests better cardiorespiratory fitness.
• Fatigue Index (z-FA): This compares predicted versus actual “mechanical cost” (PlayerLoad™). If a player generates a higher mechanical cost than anticipated for their movement, they are flagged as being in a potential state of fatigue.

3. Match Minute Prediction and Simulation

The model uses a Random Forest regression to predict match playing time based on the load history and these indices.

In practice, practitioners use this model to run “what-if” simulations:

• The staff holds all of a player’s variables (like their recent training load and fitness) constant.
• They then manipulate the Match-Day Fatigue Index to represent different scenarios: Baseline (typical fatigue), Low, Moderate, and Severe fatigue.
• The model then outputs the specific match minute (e.g., 65 min, 78 min, or 82 min) at which that specific player is expected to transition into those fatigue states.

4. Applied Decision Support

Coaches use these minute-based thresholds to make concrete, data-driven decisions:

• Starting Lineups: Determining if a player has enough “minutes in their legs” to start.
• Substitution Timing: Identifying the optimal window to sub a player off before their performance drops significantly.
• Return-to-Play: Managing the reintegration of injured players by setting safe minute limits based on their current simulated fatigue thresholds.

The research found that these thresholds are highly individual and change based on the season period, the player’s position (e.g., wingers reach fatigue earlier than centre-backs), and their injury status.

---

Moving Beyond Total Distance

The researchers found that simply looking at kilometers run is misleading. To truly know a player’s limit, you must look at Mechanical Work and High-Speed Running (HSR):

• For Coaches: A player might reach their fatigue threshold at 60 minutes in a high-pressing game, but could last 90 minutes in a low-block defensive game.
• For Players: Your “minutes” are not a fixed number; they are a “budget” of high-intensity actions. Once you spend that budget, your risk of soft-tissue injury (like hamstring strains) increases significantly.

The “Individualized Threshold” System

The paper proposes that every player has a unique “signature.” By analyzing four seasons of data, the model predicts:

• The Safe Zone: Minutes where the player can maintain tactical intensity.
• The Danger Zone: The point where sprinting speed drops and heart rate recovery slows down.
• The Substitution Window: Identifying the 5–10 minute window where a player should ideally be subbed off to prevent overtraining or injury.

Key Takeaways for the Pitch

For the Coach:

• Use Data to Validate Substitutions: Use the predicted “Fatigue Threshold” to plan substitutions before the game starts, rather than waiting for a player to look tired (by which time the injury risk has already peaked).
• Rotational Strategy: If the model shows a player only has “45 high-intensity minutes” due to mid-week fatigue, consider starting them on the bench as an “impact sub” rather than starting them and risking a breakdown.

For the Player:

• Be Honest with RPE: The model relies on “Internal Load” (how hard you feel you are working). Accurately reporting your Perceived Exertion (RPE) after training helps the model predict your match-day minutes more accurately.
• Training Consistency: The “Fitness” side of the equation only grows with consistent, high-quality training. Skipping “top-up” runs after a game where you didn’t play many minutes lowers your predicted threshold for the next match.

Note: This summary was generated with the assistance of Gemini based on the original article and additional sources, with the aim of translating the research into practical insights for coaches and practitioners.