Football is a subjective sport. 

But statistics and data are becoming more and more important. Clubs are using fewer physical scouts and relying more on data analysts when it comes to tactics. 

As the importance of statistics grows in the modern game, we want to look at key statistics to see just how useful they are when predicting the outcome of a player or team's performance.

There’s one statistic that pundits, fans and media focus on when discussing team performance.

Possession. 

The most common possession stat is calculated as the share of passes that each team had in the game, as a proportion of the total.

It’s one of the more divisive debates in the modern game. The silver bullet to some; a futile statistic to others.  

So, we wanted to settle the debate for good. 

Is possession worth looking at? 

Let’s first take a look at what others have found. 

The possession debate

An analysis of the English Premier League by Parziale and Yates in 2013 found that ball possession is an important statistic.

The study strongly correlated points earned and ball possession in a regular Premier League season. It also revealed that teams in the top four teams in the English top-flight had higher rates of possession.

Hoppe et al. (2015) found a similar relationship between total distances covered with ball possession and points accumulated in the 2012/13 German Bundesliga. 

There is one study in 2021 that looked at the relationship between ball possession and match outcome in the UEFA Champions League.

Then there was Collet’s study in 2013 which, on the surface, looked like it agreed with the studies above. Having analysed a mixture of some of the most important competitions in football, it concluded that possession was a direct predictor of match outcome since the teams with more possession shot more and scored more goals. 

But, when they accounted for home advantage and removed the strongest teams from the study, the relationship between possession and match outcome weakened. 

As it stands, there’s no clear conclusion on whether possession is a key stat for determining match outcome. 

Adding to this, the modern game has changed over the past decade. The onset of data analytics has meant that clubs employ a flexible approach with tactics by adapting to their opponent and match scenarios. This has meant that no one style of play - whether that’s possession-based or counter-attacking - sits above another. 

Looking at recent Premier League seasons, let’s take a look at whether possession is important or not.

Possession is related, not correlated

There have been some interesting instances where possession has not been an important factor at all.

Take Leicester City’s infamous Premier League winning season. 

In our analysis of their 2015/16 league win, we show that Leicester broke the mould of a Premier League champion side. 

The table below shows that Leicester were very different to the average winners over the 10 years before they won the league. Most interestingly, Leicester had just under 20% less possession than the average Premier League winnings side. 

Not only does this show that possession isn’t a determining factor of success, but it also shows that any style of play can yield positive results.

Leicester is a great example of a team that was extraordinarily clinical and executed counter-attacking football almost perfectly. Without the technical players required to play possession-based football, manager Claude Ranieri orchestrated a fast transitional play that minimised the need for long phases of possession.  

There’s no doubt this was an anomaly. But it does show that possession, even over a 38-game season, is not a tactical necessity. 

It’s worth analysing whether this trend continued into the following Premier League seasons. 

Starting with the 2022/23 season, we compare league finish with possession. If possession were a key determiner of success, the two tables should mirror each other to some extent.

The treble-winning Manchester City team tops both tables. They had the highest amount of possession (64.7) and finished first in the league with 89 points. 

But the rest of the table doesn’t show that possession is a good indicator of a team’s success. Interestingly, all three teams that were relegated that season weren’t in the bottom five teams for possession. 

Equally, Newcastle United and Manchester United both rank outside of the top four in terms of possession, but finished within the top four. 

So, although Manchester City finished top with the most possession, it’s safe to say that there’s not a clear correlation between possession and league finish.

The same pattern can be seen in the current Premier League season. 

Before Matchday 18 of the 2023-24 season (the time of writing this), the relationship between possession and wins on a game-by-game basis is negligible. Of the 168 Premier League matches that have been played, 90 have been won by the team with more possession. That’s just over 50 per cent of the games played.

There’s so much to the game other than possession that, if the only information that you knew about a particular Premier League game was the possession of the two teams, it would offer very little information about the outcome of the game. 

Some examples spring to mind. The time that West Ham had 21 per cent possession but won 1-0 against Brighton or Chelsea’s 1-0 loss to Nottingham Forest despite having 78 per cent possession. The most notable was Aston Villa’s 6-1 thrashing of Brighton with 38 per cent possession, a clear example of how goals and possession are not closely correlated. 

In the examples above, Brighton’s lack of defensive strength played a much bigger role in their loss. 

Sky Sports found the same to be true in the 2017-18 Premier League season. Their analysis found that of the first 90 matches, 35 had been won by the team with the most possession compared to 33 that were won by the side that saw less of the ball. In terms of points, it was 127 compared to 121 - so only a marginal difference. 

Even in the season before, when Leicester won the league, 137 games were won by the team that had more possession and 131 by the team that did not. 

What’s better than possession?

Look, when it comes to possession, it’s the top teams in the world that will often have more possession. But, as the analysis has shown, that’s not because it directly leads to games.

A team’s quality of chances and the extent to which they’re clinical are far better determiners of success. 

Possession is a luxury, not a necessity. 

With this, it naturally begs the question: what statistic should we look at when attempting to predict a team’s outcome? 

In recent years, expected goals (xG) has become the go-to statistic. It measures the quality of shots that a team takes to predict how many goals they are likely to score.

This comes in handy in situations where a team has far less possession than their opposition but is creating good-quality chances. The opposite is also true. A team can dominate possession but if most of their shots are from outside of the box, the likelihood is that they will score less and win less. 

For coaches, it’s better to analyse the quality of possession.

Data companies are developing new ways of making sense of numbers such as FBref’s possession-based analysis for individuals. It reveals how effective a player is in possession like how many times they carried the ball at least 10 yards or how many passes are completed that move the ball 10 yards closer to the opponent’s goal line. 

These are much clearer examples of how possession can determine a team’s likelihood of success.

It’s simple, possession itself is a useless stat. It reveals next to nothing about a team’s likelihood of success. 

As stats become more fine-tuned, we’ll see the possession stat evolve into something far more meaningful. It will help to provide more context to other key modern stats like xG and manoeuvre its way back into relevancy.