SciSports InsightSciSports
After the clock strikes midnight on January 31, most European soccer clubs will be unable to sign any new players for another six months.
The month-long winter transfer window is the only opportunity for teams to strengthen their championship bids, aid their battle against relegation, or ease an injury crisis.
January is a particularly difficult time to purchase players. Teams often do not want to sell and there is no pre-season to gradually introduce a new player to a club's tactics and philosophy. Coupled with the recent inflation in transfer fees, it's important that teams get their acquisition strategy right.
The biggest clubs employ huge scouting teams to watch matches around the world, filing detailed reports on physical, technical and mental attributes and a much greater emphasis on data analysis has been applied to soccer recruitment.
Life imitating art
But even still, it can be difficult to predict the impact of a particular player on a team – especially in a January transfer window when you hope a signing will hit the ground running. This conundrum is something Dutch analytics firm SciSports hopes to solve.
SciSports sees video games as inspiration. In Football Manager you have access to a wealth of statistics about real-life players, while FIFA offers 3D visualization of matches. In the FIFA Ultimate Team mode, gamers can see the potential impact of a signing on their team through the 'Chemistry' statistic.
Giels Brouwer founded SciSports two and a half years ago after completing his studies in the Netherlands and is influenced by the user experience and data-driven recruitment options they offer.
"I did my thesis on how FC Twente [the local football club] had improved its transfer strategy through the use of data," he tells me. "The idea [behind SciSports] was to build a real-life Football Manager using data by helping football clubs build an index on all players [to aid] the recruitment process."
Recruitment
Just like Football Manager, recruiters can search for players based on various parameters but what separates SciSports' Insight platform from other analytics tools is the way it assesses players.
While other platforms crunch statistics like completed tackles, pass accuracy and chances made, SciSports only looks at events that happen while a particular player is on the pitch. Every event within 15 seconds of a goal is given particular emphasis so it's not just assists and goals that get the glory.
"We estimate the expected outcome on the pitch. If they're expected to win 2-1 and the win 3-1, then we recalibrate which players underperformed on the pitch and overperformed," he says. "Most clubs use it to improve their recruitment policy. In the UK we have Wigan, Leeds and Huddersfield."
Clubs look at this data to see whether players are an ideal fit. For example, some clubs have identified strange patterns in the data and withdrawn from a signing because of concerns about a player's mentality.
The most effective way to use the tool, Brouwer says, is part of a more rounded scouting operation that ensures a club is spending their money in the most effective way.
"The best players will always rise," he says. "I don't think any club will need us to scout Messi or Salah. That's not where our expertise lies.
"The data is never sufficient enough to rely on it alone, but it can help flag the right type of player. It's like buying a house - you'll be able to research the house online but you'll never buy it without seeing. But at the same time, you'll never drive around a city looking at all the houses and take a risk."
SciSports InsightSciSports
Opposition analysis
Some clubs use SciSports' data to analyze the opposition. Ajax and the Belgian national team are two that benefit from the analysis of hundreds of thousands of data points to identify potential strengths and weaknesses.
In games like Football Manager, these insights are presented in bite-size snippets that are easy to understand. Brouwer says this ease of use is a clear inspiration for the ongoing development of SciSports' platform.
"We're definitely influenced by video games," he continues. "They put so many hours into their UI and if they do it so well [why shouldn't we look at it]."
It is the desire to improve opposition analysis through 3D visualization that has seen SciSports team up with analytics expert SAS.
"We were not able to build the entire infrastructure to roll it out to a number of leagues. We need a partnership with an analytics company that is slightly bigger than hours," Brouwer explains, adding that the deal will also see the two firms look at deep learning possibilities and will improve the ability to assess their own performances as well as those of the opposition.
SciSport purchases data from third-party sources to power its Insight recruitment platform and tasks itself with enriching that data. For opposition analysis, the partnership with SAS will allow SciSports to use custom datasets such as GPS, video and more subjective information such as youth department questionnaires.
The ultimate aim is to build an AI simulator that can predict an entire match: "We're not there yet, but 2-3 years we think it will be possible to watch a match in a hologram style."
The future
The insights that can be gained are improving all the time, and although Brouwer concedes that soccer is the most difficult of all sports to apply analytics to, he believes there is more progress to be made.
"[The level of insight] has definitely improved," he argues. "[For example] one of the things we see now is the measurement of goal threat. Based on the direction of the goal, we can estimate the likelihood of a goal. You can also [digitally] see a radar on the pitch and identify the spaces on the field.
"For us, deep learning is the most interesting [future development]. We have about 20 people working on deep learning and machine learning.
"[Soccer] is definitely different. It's the most complex sport in the world but I think that's the beauty of the game – you will never create a robot match."
© Soccrates Images/GettyImages Sevilla v FC Barcelona - Spanish Copa del Rey 