How stats have changed Premier League football forever

17th Aug 2013 | 08:00

How stats have changed Premier League football forever

The season has only just started, but has Mourinho already got it won?

How stats have changed football forever

It's April 30 2012 and the Etihad Stadium in Manchester is bristling with hope and anticipation. For the first time in decades, the blue half of the city sniffs glory.

The only thing obstructing it is their bitter rival, Manchester United. A win would give City a huge psychological edge in the title race, anything less would almost certainly result in the championship staying at Old Trafford. Again.

Just seconds before half time, City win a corner. David Silva steps up and delivers a wicked inswinger with Vincent Kompany rising highest to deliver a bullet header to give City a lead they would not relinquish.

City would go onto be Champions, but while Mancini and co got the plaudits, few folks mentioned the data guys...

Pitch perfect data

"Manchester City's data department analysed about 400 corners in several national leagues over seasons, and concluded that the most dangerous corner is the inswinger: the ball that swings in towards goal," author Simon Kuper wrote for Le Monde Diplomatique.

"The data team took this finding to City's manager Roberto Mancini, whose gut told him that the most dangerous corner was the outswinger. Mancini's assistant David Platt came to chat to the data analysts, and they noticed that City had begun taking inswinging corners. That season City scored 15 goals from corners, the most in the English Premier League."

In a game of small margins, this was as small as they come, but it was enough.

City's statistical enlightenment represents just one way the top football clubs are using complex data tools and armies of analysts in order to gain any slight advantage on the pitch. Stats readily furnished by companies like OptaPro and ProZone, and enhanced by powerful analytics software, are becoming king in all aspects of the game.

They're being used to analyse in-house performance, to spot vital chinks in the armour of opposition teams and pinpoint areas that should be targeted. Performance departments give huge credence to a player's numbers when recruiting new signings.

The Michu effect

Swansea City pulled off, arguably, the bargain of last summer's transfer window when they signed Spanish forward Michu for just £2m. He scored 18 goals while £50m Chelsea striker Fernando Torres scored 8. Value for money, much?

Blake Wooster a football analyst and founder of The 21st Club, an initiative that encourages clubs to quantify all aspects of their operation, told us: "Numerous scouts went to see the player, and didn't like the "look" of him. The data said he was under-valued, that his performance was high in relation to his buy-out clause value, and so it has proved..."


Simon Farrant, the Marketing Co-ordinator for OptaPro, which supplies data to Chelsea, Manchester City and the Brazilian national team said: "It's all about clubs trying to find little edges that were previously being ignored."

It's all about clubs trying to find little edges that were previously being ignored.

"Different metrics have become more important. I know a lot of clubs have talked about regaining possession in the opposition's final third being very important in contributing to the number of goals scored, but clubs who've discovered these edges don't tend to go shouting about it publicly, obviously."

Fan metrics

It's not just the clubs who're benefitting from the dramatic increase in available data. Fans also have access to astonishing volumes of information, enabling them to analyse metrics way beyond the basic stats served up by broadcasters, (admittedly hampered by time constraints and a viewer distaste for on-screen graphics) only making use of around 5 per cent of the data available from Ofcom.

One of those services is the free second-screen app Squawka, which licenses Opta data and feeds 500 million data points per game (up from 14m last season) and into its complex algorithms. All in real time no less, bringing up to the second data while the game is going on.

Squawka takes everything Opta dishes out. As a result fans can see exactly where on-ball actions like fouls, interceptions, passes, headers, duals and take-ons happen on the field.

They can see where corners, shots and through-balls are directed and which were successful. They can match players and teams up against each others in visualised, easy to understand "red is good, blue is bad" charts. All of those data points, or "on ball actions" are funnelled into create Performance Scores – a statistical, opinion-free means of rating a player's contribution or a team's display on any given day, in any given league, across any given season.


The London-based company's founder and CEO Sanjit Atwal told TechRadar: "I think it says lot about the football fan mentality that emotion usually overrides everything. In the past there hasn't been the scientific data to back up those emotions.

"We're working really hard to built the kind of metrics where football fans can have those common conversations with others. There needs to be this objectivity, these centre-points where you can actually have a proper conversation.

There could be a fundamental shift in the way the sport is played using this data.

"If people are having an argument with a friend about football and one guy says "your player is" rubbish, we want to give them the ability to say "your player's rubbish and here's why" we want to fuel that "because" by giving them all the information they need, visualised.

"Level of analysis is only getting better. If we achieve what we have the potential to not only will we have a great second screen platform for football fans in real time. I also think there could be a fundamental shift in the way the sport is played and managed using this data."

Who will win the Premier League? Stats explain all

The increasing use of stats in the beautiful game has led to inevitable comparisons with the 'Moneyball' revolution in baseball where stats such as a hitter's 'on-base percentage' supplanted traditional performance metrics. Is there a killer stat out there for football that could redefine how we value players and judge their performance?

"No, is the short answer," said Opta's Simon Farrant. "There's massive differences between the two sports. Baseball is all about individual actions. It's pitcher against batter, then the play finishes and you move onto the next one. Football is an 'invasion' sport, there's 11 players on each team moving around independently of each other."

His colleague Simon Banoub, Opta's Marketing Director, agrees: "There are going to be different shortcuts over time to get to a quantifying players contribution, but the killer stat between winning and losing a game, I think football is too fluid for that to be honest. However, a lot of the analyst community are looking into that kind of thing all the time."


The general opinion appears to be that football is home to too many intangibles. Too many things that can happen to offset the individual outcomes of a player's action, for example a winger can't control whether his centre forward nods home that pinpoint cross that leaves the goalkeeper in no-mans land.

The most important things are the hardest to measure.

However, sports analyst Blake Wooster thinks there is a path to statistical nirvana in the beautiful game. "It's harder in football [to quantify performance] given the team dynamic, but not impossible," he said:

"You just need to apply contextual intelligence to the problem, and also accept that there are some things that are inherently hard to measure - the so called 'intangibles'.

"The paradox in football is that some of the most important things are the hardest to measure; creativity, game intelligences, spacial awareness, etc... Hard, but not impossible."


However, while clubs seek any advantage they can on and off the field, it's also worth considering whether there's a desire for fans to turn the beautiful game into a collection of mathematical equation.

Not everyone wants their intangibles quantified. Supporters who wish to seek out data can do so through services like Squawka, which work exceptionally well. For everyone else, increasing the desire on stats may rely on giving them entertainment value and context.

Richard Ayers, former head of digital at Manchester City and CEO of the Seven League digital media firm believes in 'datatainment' – taking cold hard stats and creating entertainment opportunities through apps and games in order to increase engagement.

"Football is such a globally popular game that I think a lot of people involved in it don't see why or how it can be enhanced by the use of data," he told us. "There is a degree of tribalism here. Data is for 'statos,' the thinking goes. Other sports use data as a key to unlock the mysteries of their game, as a way of learning and understanding more.

Our projections are compiled using a vast amount of data.

"Occasionally genuine analysis is achieved using data, but far less often than you'd think when used around football. Much less than NFL, or MLB, or even cricket where data analysis is an accepted complement to the game.

"Partially this is because football relies on 'flow' and what architects call parametric data. For example, the interrelation of all the constituent parts, whereas the other games are broken into plays, moments, individual contribution much more.


When it comes to the use of data by people communicating about the game, the essential component, context, is usually what is missing. The data itself is useless unless you give it context."

Predicting the winner

The power of stats in the modern game will be put to the test this season thanks to a spot of predictive analysis. Bloomberg Sports, using Opta data and its own analysis tools, have determined that Jose Mourinho's Chelsea have a 35.1 per cent chance of winning the title come May, finishing with 81.7 points.

Bloomberg's Simon Miller says: "Our projections are compiled using a vast amount of data - it is all wrapped into Match Analysis.

Historic performance is a key factor - this goes back through 3 years of data including the very latest action. For each of the 380 matches played in EPL there are between 1000 and 1500 pieces of data to analyze. We're also taking into account the influence of new managers and players signed.

"We have a lot of confidence in our projections. We are using advanced analytical that Bloomberg has developed in the world of finance over the past 30 years. We see the league being very tight between Chelsea and the two Manchester clubs. We give Chelsea the edge - the return of Jose Mourinho and his signings are part of the minor favoritism we give Chelsea."

However, before John Terry quickly changes from suit to kit (shinpads 'n' all) to lift another trophy, here's another stat for you from Squawka: "The Premier League has never been won by a team that finished outside the top three in the last season.

"That's just never happened, so that's a really good jumping off point to begin predictive analysis right there," countered Atwal.

Something's got to give, but it won't be the numbers. They can't get injured, request a transfer, lose form or misplace a pass. Everything else is fallible, well everything except Jose Mourinho.

football Premier League big data stats
Share this Article

Most Popular

Edition: UK
TopView classic version