Scouting smarter - Machine Learning 1 - 0 The Hunch

Wycombe Wanderers want a young, left footed roaming midfielder who has set piece quality and is regularly providing assists from outside the area, who shows leadership qualities and potential to be grown into the new face of the team.

In years past, this kind of requirement would go to the scouting team who would then scour the lower leagues and youth academies to find a slew of potential targets. These would then be observed, both with previous film of them playing and also attending live games to take notes on performance and potential.

After completing this analysis the scouting team will end up with a selection of players who could fill the position and further observations may take place. Once the final shortlist of players has been compiled, the scouting work is done, and it's on to next position.

The whole process is longwinded and fraught with risk, the scouts don't necessarily have the luxury of watching dozens of games for each potential player and could catch the next Cristiano Ronaldo on an off-day. There are also travel issues, where a great prospect might be playing for Alaska City FC, but even before the current travel restrictions, the 40 hour trip one-way may just push that player outside the reach of the club.

So how can technology help? Well, in the upper echelons of the game clubs are already accustomed to the use and deployment of data scientists to bring logic, mathematics and an objective, scientific view to proceedings. Players wear fitness trackers with data being harvested for training sessions, games and even injury rehabilitation to track exactly how each player is progressing and tailor their training regime and skills to focus on accordingly.

The mass of data that can be analysed and machine learning used to build mathematical models indicating trends in everything from the likelihood of a set piece from a particular angle resulting in a goal to the optimum number of minutes a player should be on the pitch to avoid injury and improve recovery.

With football beginning to embrace technology in more sections of the sport the scouting of new talent seems a potential area to make fast progress - and this is already in play. IBM's "JAAI SCOUT" project [now spun out into "Scoutastic"] initially worked by digitising and then training the system on previous scouting reports to identify patterns are particular qualities that should be looked for in new talent. German Bundasliga team Weder Bremen have been a partner in the development of this technology and are changing their process to fit with the advancing technology.

Previous scouting information, while a vital barometer for what works at an individual club isn't great at finding those new players who have the attributes the club requires, which is where video analytics, alongside the machine learning technology can step in and take the pain out of scouting. Using footage from every source available, teams could insert the attributes they require, much like in the popular computer game series Football Manager and the list of prospects would be automatically generated with a high degree of accuracy, removing the need for the ultimately wasted trips to training grounds far and wide.

This level of confidence would only improve with time and use of the system, where the algorithm could be tweaked to iron out any initial kinks. The time, expense [let alone environmental impact] reduction would give the scouting team the ability to visit more of the high-probability targets and compile the shortlist in a much more efficient manner.

As with all systems that are to flourish, while they must speak to a current need and solve a real-world problem they must also serve alongside the experts in the field and not look to replace. The scouts are at the top of their game professionally and can make a judgement on a player within the first few minutes of seeing them with a ball, so any solution that arises should give the power to the scouts and make sure they are concentrating on the proper player evaluation rather than working out their travel schedule.