Predicting the training load of team sport practice sessions

Kenny McMillan

October 14, 2022

Data Science

TLDR: Drill Planner is now available in Orreco Te@m - a data analytics and risk management platform for pro sports.

It is common nowadays for the performance staff of professional sports teams to use global positioning system (GPS) tracking devices to monitor the training load of players. Performance staff can use the data output from these devices to quantify the total workload of a training session and the workload of each individual drill to monitor each player's short -and long-term loads.

Performance staff typically deliver training reports to the coaches once the GPS data is processed post-session. The reports are typically generated using reporting tools built into the GPS software or through a data export into business intelligence platforms such as Tableau or Microsoft Power BI.

What is far less common, however, is the reporting of the predicted training loads PRIOR to the prescribed training session. It is all very well giving nicely formatted and visual reports to coaches after the training session has finished, but it is far more impactful to report the PREDICTED training loads BEFORE the session has even started.

When designing a training session, the ability to predict the expected workload demands of the session is clearly beneficial as part of routine player load monitoring, as the coaching and performance staff can check if the workload for the planned session is appropriate for that given day. Quantification of the expected workload of a training session can also prompt the soccer coach to alter the training session to attain the required workload.

While working in British professional soccer a few years back, I recognised the need to predict the workload of training sessions to optimise the players' training and mitigate fatigue and injury. Unfortunately, my attempts to predict and report future training loads to the coaching staff proved futile. I was restricted by clunky, slow Microsoft Excel worksheets, elementary calculations, and my (very) limited statistical knowledge at the time. I found that I could not predict future training loads with any worthwhile accuracy.

When completing my Master's Degree in Data Analytics in 2019, I decided to revisit the training load prediction scenario for my final year project. Using historical training data collected from a senior professional soccer team over several seasons, I applied the statistical programming language R to create appropriate statistical and machine learning models to predict the training loads of prescribed training sessions. These predictive models were then deployed in a user-friendly R Shiny application to enable the end user to derive quick and actionable insights.

An early iteration of the Training Drill Planner where various predictive models were deployed and fine-tuned.

Embraced by Orreco

Brian Moore, CEO and founder of Orreco, immediately recognised the potential of the Drill Planner. Their world-class team of statisticians and data scientists optimised the predictive power of the system and Drill Planner is now available as a module in Orreco's recently released Te@m platform, a data analytics and risk management platform for pro sports.

Orreco's world-class team of statisticians and data scientists optimised the predictive power of the Drill Planner.

Supporting decision-making

In my experience, soccer coaches are highly proficient at prescribing training sessions for technique,  tactics, and match preparation but are usually unaware of the physical load of their sessions. I can recall many examples of seeing this first-hand on the training ground. For instance, at one professional club I used to work at, a crossing and finishing drill was carried out in the training session two days before a competitive league match. What was an excellent technical football drill also imposed a high number of accelerations and decelerations, way above the typical mechanical load values for two days before match-play. This unaccustomed mechanical load led to some players reporting above-normal muscle soreness the following day. On the match day, one of the strikers who had experienced muscle soreness resulting from this training drill pulled up abruptly in the first half with a tight hamstring when making a sprint. He had sustained a Grade 1 hamstring strain that kept him out of the team for a few crucial upcoming matches.

A powerful tool for soccer coaches and support staff

The Drill Planner quantifies the expected workload of a training session, which can prompt the alteration of the training session to attain the required workload. It also highlights certain combinations of training drills that players find difficult or easy. Performance Support staff and coaches can use these insights to manage injury risk and prescribe individualised load adjustment. Predicting the workload of the session helps in selecting the optimal choice, duration and number of drills to induce the required training stimulus.

The Drill Planner is also a great tool for coaching educational purposes, empowering the coach to increase their understanding of the physical loads of the individual training drills and sessions they deliver.

A synergistic approach

As a stand-alone solution, the Drill Planner is a really effective tool for optimising performance and reducing injury risk, and it is great to see numerous soccer teams already using it. But what excites me most is its integration into the Orreco Te@m platform. This platform enables the amalgamation of multiple data sources arranged into modules on the platform, and the collective of these modules is powerful.

The Drill Planner becomes even more impactful when its insights are complemented with jump data, sleep monitoring, movement signature analysis, biomarker data, menstrual cycle data and performance nutrition to give a more complete profile for risk management and training optimisation.

For example, biomarker information may indicate that your star striker’s FORD and FORT levels are outside the player's adaptive range. Sleep monitoring data from an Oura ring fed automatically into the Te@m platform may reveal that the player had a lousy sleep. Now, suppose the Drill Planner predicts the load for the upcoming training session to be abnormally intense. In this scenario, performance staff can use all these combined facts for optimal decision-making. In this example, the best decision may be for the player to participate in selected drills of the training session or even miss the training session altogether. Of course, it would be wise to inform the coaches that the prescribed training session is more intense than usual and that the workload resulting from the choice, duration and number of the drills prescribed may put other players at risk of fatigue and injury due to the excessive overload.

Te@m: a solution for data analytics and risk management in pro sport. The unified platform connects every player with trainers, medical teams and executives, helping them to truly understand the demands the players are experiencing.

In female soccer, combining Drill Planner with menstrual cycle tracking tools like Orreco’s FitrWoman platform can generate powerful insights to maximise player health and performance. Armed with information on each athlete’s menstrual status, training programs can be optimised to account for important considerations within each phase of the cycle. It’s all about being proactive and prepared – Drill Planner and FitrWoman together provide performance staff with an effective solution to managing training load, player by player, cycle by cycle.


There is a fine balance between optimising training load to maximise performance and minimising injury. If you provide performance support in team sports, are you providing your coaches with the predicted training load of their prescribed sessions? If you are not, maybe it is time to start doing so.

Delivering a physical workload report after a session is too late if the resultant load is inappropriate. As I have observed many times during my work in professional soccer, the delivered training session can be a great one in terms of soccer technique, tactics, and game preparation, but it may have imposed an unsuitable training load at an inappropriate time.

It is time to start using your GPS data prospectively rather than retrospectively to support decision-making, reduce injury risk and enhance the performance and welfare of professional players.

If you would like to learn more about how Orreco is using statistical modelling and machine learning correctly to help identify injury risk in professional sports, check out the recent post on the Orreco Motion Data Collective.

Have a question?
Get in touch
Other Stories From The Blog
See All
Sign Up to our Mailing List
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
By signing up you agree to our Privacy Policy