Members Login

Join ASD

Analytics in Football – the past, the present and the future

pexels-pixabay-46792-scaled-aspect-ratio-1251-605

6 Min Read

Feb 04 2019

By Professor Ian McHale

It is nearly 20 years since the Oakland Athletics employed Paul dePodesta, and Billy Beane started to use analytics to gain an edge in Major League Baseball. Outside of sport, we have since entered the era of big data, and mega-corporations like Google, Amazon, and Facebook boast about using analytics in all aspects of their businesses. Elsewhere in sport, the success of Team GB and Team Sky Cycling in recent Olympic Games and at the Tour de France has been attributed to Sir Dave Brailsford’s adoption of analytics in search for “marginal gains”; whilst Formula 1 teams employ armies of statisticians to determine which race strategy maximises the expected finish position for their drivers.

In this article, we take a look at the past, the present and the future of analytics in football.

 

The Past

It is probably fair to say that the relationship between analytics and football has historically been a poor one. In the 1960s, a statistician called Charles Reep collected data on passing sequences and wrongly concluded that short passing sequences (the longball) were more likely to result in goal scoring opportunities than sequences consisting of lots of passes. This is how the infamous longball was born. Jonathan Wilson argues in Inverting the Pyramid that this error had negative consequences in the way football was played in England for 50 years! This is certainly not a success story for analytics, and did not represent a good start to the relationship between analytics and football. It should in fact act as a lesson to those making decisions using the results of analytics “can the results be trusted?”.

 

The Present

It has taken a long time for any significant improvement in the relationship between analytics and football to be made. But, over the last decade or so, perhaps driven by the analytics revolutions happening elsewhere, there has been a gradual realisation within football that analytics and data can contribute to a football team’s success. In tandem with ever improving data, the field of football analytics has developed. Clubs now employ analysts to study, for example, passing and shooting statistics, in the hope of unlocking the key to future success. The widespread appreciation of the concept of expected goals suggests that football has, to some extent at least, begun to appreciate that analytics has something to offer!

The journey towards a happy marriage between football and analytics is far from complete. If truth be told, football and analytics have been on a couple of dates, and whilst analytics thinks football is very attractive and has high hopes of living a long and happy life together, maybe football has some serious doubts about the long-term outlook for the relationship.

Football is a complicated game. Unlike baseball in which we could say there is one input (the pitcher), one variable (the batter), and one output (the outcome of the pitch), football is a game of many inputs and outputs, with dozens of variables. Add this to the fact that all of the variables are continually running around a pitch, and perhaps you start to see the problem – the nature of football means that analytics in football is extremely hard. And this complexity, and the level of expertise of people charged with producing meaningful analytics out of this complexity, means that analysis is perhaps not as insightful as it could be. Analytics in football is currently far from being the finished article, and strong elements of doubt remain as to whether analytics will ever really be useful in football.

 

The Future

To me, there is no doubt about it, analytics can be used in football to help make better decisions. It is certainly not going to replace decision makers, but if used correctly, it will improve the success rate of decisions. If Google set its expert analysts the task of buying, and running a Premier League football club, I have no doubt that within a few years, that club would be seen to over-perform relative to spend.

The difference between an imaginary Google FC and a typical club at the moment is, in my opinion, an appreciation of what analytics actually is, and what it can do. Analytics is the process of using data to making better decisions, and crucially, analytics should have a positive impact on performance, and more importantly future performance. Analytics is not collating simple summary statistics of passes complete in the final third with no appreciation of whether knowing this is indicative of future performance, or whether this information can improve decision making. If an analytical technique does not have a demonstrable positive impact on the performance of Amazon, for example, by improving the customer’s experience, do you think Amazon continue to use that analytical technique? Of course not. And Amazon will test for this impact on performance. Has anyone reading this article ever seen a test of whether using expected goals to assess a player’s or a team’s performances improves decision making and future performance of the team? My guess is probably not – and this is where analytics in football currently falls woefully short. Various pieces of analytics are bandied around with an assumption that the analysis is good. But without appropriate testing, we should never accept a piece of analytics as good, no matter how intuitive or “clever” it might seem. This is part of the scientific process.

I expect the future for clubs can take one of two paths:

  • Some clubs will fail to appreciate what analytics is, fail to realise that it can be used to aid decision making, and make non-optimal decisions. These clubs’ success-to-spend ratios will decrease as recruitment decisions (managerial and player) will be less successful than for a second group of clubs.
  • Other clubs will persevere with analytics, and identify people, methods and tools that contribute to making better decisions, decisions backed by evidence. For these clubs, success-to-spend ratios will increase compared to the first group of clubs.

The clubs taking the first path will fall behind the others but will eventually realise the source of their disadvantage. Unfortunately, it will take several years, and resources to catch up to the clubs choosing the second path.

Sporting Directors should ask themselves: “which path should I take?”.

 

We use cookies on our website to ensure that we give you the best possible experience.