Data Analysis Revolution in Football

During the 2014 FIFA World Cup held in Brasil, Iran and Nigeria played out their opening match to a scoreless draw. Pre-match analysis and pundits had been tipping Nigeria to win the match but it wasn't so.

FIFA World Cup 2014 Iran vs Nigeria Statistics

Image Credit: FIFA.com

Post match data and statistics that is usually tracked and displayed by media companies (online, television and print) suggested that Nigeria should have easily beaten Iran as the Super Eagles dominated play in all attack related statistics.

But, as early as the end of the first half, other data suggested that Nigeria probably need to change tactics (which they didn't in the second half) if they were to score a goal.

Only until the 87th minute in this match did Nigeria have a succesfull cross (26 failed crosses before that, i.e. cross didn't find a Nigerian player) into the 18-yard area.

FIFA World Cup 2014 Iran vs Nigeria FourFourTwo Match Zone

Image Credit: FourFourTwo

Throughout all of the first 45 minutes, Nigeria attempted 11 crosses from both sides of the pitch and none found a Nigerian player. Add to this that throughout the whole match Nigeria had a mere four shots on target and only one of those from inside the 18 yard area, thanks to data analysis it becomes quite clear that the Super Eagles tactics against Iran relied heavily on crosses.

What did Nigeria do at half time when it was already clear that their crosses strategy is not working and that they haven't even attempted more direct shots at goal, nothing (at least the data tells us that).

Eight minutes into the second half Shola Ameobi (a slightly more central playing attacker based on the heat maps) was brought on for Victor Moses (one of the wingers and culprit of failed crosses) and Ameobi immediately showed his worth with two shots on target.

FIFA World Cup 2014 Iran vs Nigeria Statistics

Image Credit: FIFA.com

That was a big improvement already but soon thereafter the crosses strategy persisted with match data analytics showing us that there were more crosses made by Nigeria in the second 45 minutes (18 crosses) and only 2 were succesfull with the first succesfull cross coming in at the 87th minute.

You might say in hind sight it is easy to say this, but as illustrated above, the data was already available at half time and as the match progressed. The question then becomes, did Nigeria's coaching team have access or use of such analytics tools or did they ignore them?

Pioneers of Football Data Analytics

Although the Nigerian Football Association provided us with no answer to whether they use football data analytics tool/s, a quick look at Opta (one of the leading sports data analytics companies) suggests that Nigeria (or any African football club or body) doesn't use them. Fair enough, Opta also declined to comment along with other sport analytics companies.

Although I digress, slightly, the point is were the Super Eagles using such data during the match, they would have switched tactics and attempted something else.

Opta has been in the football data analytics (and now covers cricket, rugby, golf and tennis too) for many years dating back to the mid-1990s when it was first associated with the English Premier League through its Opta Index service.

Since those days of their player and team ranking service (Opta Index) they have since started collecting more data sets during a LIVE match and updating their database and analytics platform. Not only is this data available to football coaches during a match but Opta now has various platforms and clients in vaious industries.

Betting - where they provide live match data and pre match data to assist bookmakers in setting odds and for bookmakers to supply this data as an added service to their clients.

Opta Betting

Image Credit: Opta Sports

Broadcasters - to make the match viewing more insightful and interesting, Opta provides some broadcasters with data analysis of the matches to keep viewers hooked on the station for longer pre-match, during half time and post match. One of the benefits of this for broadcasters is likely more Ad revenue.

Opta Sports

Image Credit: Opta Sports

Football Teams - more importantly and notably, Opta counts among its clients some of the top footballing teams and associations (other sports too like rugby and cricket) in Europe which could be part of the reason why some of the top football is currently being played in Europe and by European Nations.

Opta Sports

Image Credit: Opta Sports

Some of their other clients are in the print, online and mobile publishing industries.

Another company that is innovating and pioneering in the field of football data analytics is SAP.

Not only does the German software company, SAP, develop a Match Insights program whose analytics are viewed and analysed on touch screens but they have extended football match data analysis to the training field as well.

Using their SAP HANA platform and wearble technology (player bands, sensors built into shirts and shin pads as well as sensors in the ball) the software company helps the German Football Team along with notable German football teams such as Bayern Munich and TSG Hoffenheim track player data during training.

Data from training sessions combined with match data definetly goes a long way to increasing any team's chances of winning matches and optimising their performances.

Transfer Market

In their book titled Soccernomics, Simon Kuper and Stefan Szymanski go into detail and give examples of how quitely data analysis is slowly changing and informing transfer market decisions by football clubs and coaches.

The two notable examples, which are similar and illustrate the importance of digging deeper into data analysis, are of how Sir Alex Ferguson decided to sell Jaap Stam and later admit (somewhat) he was wrong and how AC Milan decided to keep (and play) Paolo Maldini as captain until he retired at the age of 41.

Soccernomics: Why England Loses, Why Spain, Germany, and Brazil Win, and Why the U.S., Japan, Australia-and Even Iraq-Are Destined to Become the Kings of the World's Most Popular Sport

In football data too there is noise (vs signal that is a call to action), and this data noise is what partly led to Sir Alex Ferguson selling one of his best defenders at the time, Jaap Stam, in 2001 for a sum of £16.5 million. The price tag alone was good enough for a 29 year old defender. But what made Ferguson sell Stam?

In the book, Ferguson is quoted saying that he looked at Stam's data over the years and picked up that his tackles per game were dropping with every season that passes. This was a misunderstanding or misinterpretation of the data as Stam went on to have succesfull football seasons playing for Lazio, AC Milan and Ajax Amsterdam before retiring.

What the data wasn't telling Sir Alex Ferguson, and perhaps he should have looked at other data, was that as Stam grew experienced with every season that passed so did his positioning thus resulting in him making more interceptions and not needing to tackle players.

Using data that Ferguson should have used at Manchester United before selling Stam, AC Milan saw that Maldini's tackles per game were dropping but they looked elsewhere and picked up that his positioning had been improving resulting in him making fewer tackles thus continuing to play and be captain of AC Milan until he retired at 41 years old.

Such data analysis does influence when and who a team buys in the transfer market and for how much, but given that some of the football clubs are bank rolled by some of the world's billionaires there's still a lot of ego in the transfer market. This presents an opportunity for clubs on tight budgets to unearth players that other clubs overlook using data analysis.

Opposition Scouting

Most of the top tier league teams in Europe and South America have data analysts and data scientists on their payroll whose sole job is to crunch player and match data. This doesn't only apply to their team's data but to the opposition team's and player's data.

They are tasked with coming up with answers to questions the coach might want to know such as a specific opposition player's pass completion rate in final third of the field (if high, the coach might want two players marking that player), pace of sprints for each attacking opposition player (if your defenders are considerably slower than the opposition's attackers, you have a big problem).

Data like this informs decisions on how a coach should setup his players and tactics for an upcoming match. Added to this, clubs that are clients of sports data analysis companies such as Opta regularly receive reports such as the follwoing:

Opta Sports Image Credit: Opta Sports

Opta Sports Image Credit: Opta Sports

It's not everything

Admittedly, data analysis is not the only thing that will win a team trophies and bring success. Many factors play a role in that, and for that I would suggest reading Soccernomics.

On its own, data analysis doesn't win trophies but it does give a team an adavantage (can't quantify but would say a 5% to 20% advantage) over a team that doesn't use data analysis.

Which begs the question why aren't we seeing a greater use of this on the continent?

Although African countries are much poorer than European ones, we continuously export some of the world's top footballers to Europe who get their skills refined in Europe yet when they re-assemble (some African National Teams field a full team of Europe based players) as a national team they fail quite dismally.

Opta and similar services don't come cheap and on-pitch player tracking technology and cameras as used by some football data analysis companies is expensive, perhaps this is also a wake up call to African developers and (hardware) makers to look at an African developed solution that can be used accross various football tier levels at a much lower cost.

There are already other organisations looking at the use of sensors and data analysis for football. One such example is WSO2, whose Director of Research, Srinath Perera, did a presentation in 2014 titled "Analyzing a Soccer Game with WSO2 CEP".

In the presentation Perera discusses and demonstrates how they used Complex Event Processing (CEP) and MapReduce-based technologies to track, process and re-enact data from a soccer match in real-time while at the same time producing useful analysis.

Perhaps sensors present a cheaper and more accurate alternative to having 22 cameras installed at each stadium tracking each player but then there's the hurdle that is FIFA and the national Football Associations and their policies of not allowing technology to be used.

Imagine how sensors could eliminate debate around whether a player is offside or not?

One thing is (almost) certain though, until African National Teams use data analysis, they won't be winning a FIFA World Cup any time soon.

Cover Image Credit: Guts Gaming

Comments