Do I Trust My Eyes or Stats When Assessing a Football Player?


Towards the end of last season, I came across the CIES Football Observatory’s rankings of the best players from the top 5 leagues in Europe. Dejan Lovren was rated the ninth-best centre-back in Europe, only behind Nicolas Otamendi in the Premier League. I wasn’t too far from putting my head through a wall when I read that. Similarly, Manchester City fans would have been bemused to see Otamendi at fourth in the list.

I could see how Lovren made the cut. After all, he had also made the list two years previously in twelfth place, with CIES then deeming him to be the best in the Premier League. They clearly like him a lot. And he has the numbers to back them up. You just need to go and check Opta or WhoScored stats and he scores very well on most defensive parameters. He led Liverpool defenders last season on interceptions, clearances, blocks, aerial duels won, passes and long balls played per game.

Those are the attributes of a wonderful modern ball playing centre-back. One must be strong, quick and technically good with the ball to fulfil those attributes. Why then did I feel that Lovren wasn’t close to being Liverpool’s best centre-back last season, let alone be one of the league’s best? For someone who watched every game Liverpool played over the year, he was a mistake waiting to happen. Joel Matip was the only centre-back in the side who provided me with a semblance of comfort through the course of the season.

A couple of weeks before the CIES report came out, I wrote a piece on Liverpool’s defensive woes. I analysed every goal that Liverpool had conceded until that point during the season and in addition to the primary errors that led to goals (which are captured and reported), looked for other errors that had led to these goals. These could have been poor clearances, balls played behind the defender for a forward to run onto which weren’t reached, crosses or corners a player was beaten to, free kicks conceded, deflections and the ball being given away to the opposition to score. If one were to look at only primary errors, Lovren made just one that led to a goal, a horrible miskick inside the box in a game against Crystal Palace. But if we include those errors I mentioned above, he was culpable in a team high 10 goals conceded. The big revelation was that he contributed to 7 goals conceded via a corner or a cross, the most by any player. Between the three other centre-backs in the squad, they conceded two.

Now, WhoScored lists aerial duels to be one of Lovren’s strengths, the other being passing. If we looked purely at the percentage of headers he wins, 65%, it all makes sense. He wins a lot of headers, and he’s comfortably ahead of his team mates on this parameter. What it doesn’t say is that in those 35% of headers he lost, he contributed to thrice as many goals as all his fellow CBs did, combined.

I am sure there are data scientists out there deriving more out of statistics from matches, with all football teams probably employing a few, but the way conclusions are arrived at about players in the public domain leave a lot to be desired. In baseball, they analyse late inning pressure situations, with player ratings for how they perform in clutch situations. I am sure we will have something similar soon in football. A stat on clutch headers lost would have revealed that Lovren, while being a good natural header of the ball, doesn’t quite win them in pressure situations.

In Germany, the company IMPECT works with the football league and clubs on two stats, Packing Rate and Impect. They look at the number of defenders in a player’s way before and after a certain play. The number of opponents outplayed during a game is called ‘Packing Rate’. They also determine that beating defenders was more valuable than a defender getting the ball beyond pressing strikers. The number of deep defenders taken out is called ‘Impect’, named after the company.

The inferences are not perfect, just like with the ones that are made with the other stats out there, but it does suggest that more people have started thinking about statistics in a manner more reflective of what actually transpires on the football field. In isolation, the stats available in the public domain about players do not provide a good picture about how good a player is, but only about how good a player could become.

A good example of a player who excited many with his stats is Jesus Navas at Manchester City. When he arrived in the Premier League, he had just delivered the most accurate crosses and created the most chances from open play for any player in La Liga. A disappointing stint later, it looks like he is on his way back to La Liga. Dejan Lovren and Nicolas Otamendi look regal on the eye for 90% of every match. The remaining 10% will give you the shivers.

If there is more emphasis on the usefulness of player attributes across situations, formations and leagues, while allowing for the dynamism of a football match, where at least 20 players are constantly on the move, statistics could one day prove indispensable to football teams. Until then, managers and clubs would probably use them to a certain extent, with more focus on scouting networks, because the eyes will always tell you more about a player than stats will, at least for now!


  1. Hi Rahim,

    This is very valid argument. I was drawing a parallel to data analytics at work. The analysis though sound scientifically were called out by our client as it did not reflect the real picture on the field. Having a good understanding of how football works on the ground is needed in addition to stats which will show the real picture.


Please enter your comment!
Please enter your name here