Analytics In The MLS Lacks Maturity

The perception around the league is improving, but data use is lagging behind


Major League Soccer has experienced a tremendous surge in interest over the last few years. It has seen double digit growth in season tickets sales, attendance, and local sponsorship, while the number of games televised has also more than trebled - up from ninety-eight night games in 2014 to 340 games in 2015. Every regular season MLS match-up was broadcast around the world, and total gross viewership has been reported at 30 million - a 50% increase on 2013 and the largest ever audience in MLS history.

There are a number of factors behind this success, and data analytics has played a starring role. It has been one of the primary tools in driving fan engagement, helping clubs to understand consumers better so they can personalize marketing materials, push out appropriate digital content around the games, and even enhance the experience for those in stadium. However, while these are all vital for driving the league’s popularity, the main reason behind the growth will always be the quality of the soccer on display, and analytics is still yet to fully mature in this area.

That’s not to say that analytics don’t play a valuable role in improving performance. In terms of scouting and training, teams like New England Revolution and Sporting Kansas City - both MLS cup finalists in the last two seasons - have utilized it to great effect. The MLS’s salary cap and the limits around free agency that set them apart from leagues around the world, are two major obstacles in terms of recruiting the best talent, and the primary strength of analytics is in selecting players who are undervalued and display potential. Portland Timbers, for example, have hired Harvard University sophomore Brendan Kent to check the data around players they are targeting to ensure that they’re the right fit. Kent explained some of these metrics to ‘For example, for a No. 10 I may be looking at metrics like expected goals assisted, among others, that we deem relevant to that role. The context of the stats is also very important, so I'm often using tools that allow me to tease out the context of a certain player's stats. This means taking into consideration location on the field and scenario, among other contextual factors, when analyzing the [data].’

In terms of individual fitness, the data produced by the plethora of wearables that players are now saddled with is also incredibly valuable. It helps to establish what exercises will best enhance a player physically, pinpointing the areas where they are lacking so coaches can identify any exercises that would be beneficial, or could actually be causing harm to a player.

Analytics can also show the type of training sessions that teams should be doing by establishing which are most useful for developing your model of play. And this is the key point, and where it is easy to get misled by the data. Analytics should be used to help teams play in the style they want to play in by aiding training and scouting the best players to suit the model, but they should not be used to determine the style of play itself. There are simply too many variables on a pitch for it to be accurate. Even pitch sizes and surfaces vary from stadium to stadium.

The way it works at the moment, for the most part, is by aggregating the data available. An average can show how the best teams play, and it is tempting for everyone to simply try and emulate this, but this ultimately results in everyone trying to squeeze through the same door at once, so to speak. For example, in previous seasons the data will have shown that a team who take the most shots and have the most possession will win. Leicester City exposed the folly of this in the English Premier League, winning by a clear 10 points despite coming 10th in the league for shots taken, and 18th for possession. The problem with averages is also that they don’t offer any insights as to why teams are not performing as expected, only that they’ll likely come back to normal. There are so many variables though, that there could be any number of reasons for this, most of which would not be revealed until you watch the videos back for yourself.

New England Revolution’s team president Brian Bilello told that ‘I think the trickiest thing for soccer right now, because the analytics are complicated and we are not a statistical sport like baseball, is really bridging that gap between the analytics and the head coach and everyone in between to make sure you are looking at things that make sense to the coaching staff that they can use.’ At the moment, data analytics is still not mature enough in soccer to do away with coaches. It is still a tool for supporting performance, not driving it. How much this changes in years to come remains unclear. 


Image: Photo Works / 

University lecture small

Read next:

How Are Higher Education Institutions Using Analytics?