Sergio Agüero’s 93rd minute winner for Manchester City against QPR in the final game of the 2011/2012 was not only the goal that won them the Premier League but also the 93rd goal scored by City in a remarkable season that saw the blue half of Manchester celebrate a league success for the first time in over 40 years. Fast forward 12 months and there was a different story. Not only did Manchester United cruise to the title by 11 points at the expense of City but the City strikers only managed to score 66 goals all season – a 30% decrease! During the off season, new manager Manual Pellegrini looked to solve this issue by signing Álvaro Negredo from Sevilla. Negredo had finished the 2011/2012 season with 25 goals and thus the transfer made perfect sense. Or did it?
Over the last few weeks, across the sporting world, a lot has been made of the issue of goals vs expected goals and we have seen many different articles addressing this issue. Each one of those articles offers up a definition of the term ‘expected goal’ with a formula on how to measure the scoring ability of any given player. At SportsMatrix we like to be involved in all aspects of football statistics so this motivated us to see what we could offer to the debate.
When analyzing a game, we assigned the shots taken in that game with a scoring chance ranking. What we mean by this is that whenever a player takes a shot on goal, that shot is assigned a specific scoring difficulty level. We determined the value of each difficulty level based on a number of different parameters. Those include looking at the distance from goal that the shot was taken, the positioning of the defenders and goalkeeper as well as the angle from where the shot was taken amongst others. All of these parameters will influence the difficulty of the shot. For example, the highest scoring chance would be taking a shot into an empty net whilst a low scoring chance shot would be assigned based on the distance from goal that the shot was taken as well as the positioning of the defenders and keeper.
Once we have separated the various shots into the various scoring chance levels, we divide this with the number of goals actually scored in those levels and we can see the number of goals that the average player was expected to score in that level. For example, if the average player scores 10 goals from 20 shots in a medium scoring chance level and we focus on a striker who only scored 8 goals in that advantage level, then his expected goals are 10 and thus the difference is -2. On the flip side, if a player manages to score 12 goals from this level, then his difference is actually +2. In order to work out the correct goal from shot percentage from each of the various scoring chance levels, we looked at over 130,000 shots taken in games across Europe during seasons 2011/12 and 2012/13. It is important to note that the goals analyzed do not include own goals or goals scored when a player tried to cross the ball and it landed up in the back of the net.
In the table below you can see the top 10 clubs as sorted by the difference between goals scored and goals expected. No surprises with the clubs in 1st and 2nd place. Following their poor scoring season, Manchester City found themselves in 110th place having failed to score on 7.2 less than the average team would have done when in a similar position.
As stated above, City turned to Sevilla striker Álvaro Negredo to solve their scoring issues. Interestingly, the Spanish marksman, despite scoring 25 goals last season was 4.4 goals short of what the average player would have scored being in similar positions to what Negredo had been in.
Does this mean that Negredo is not as prolific as City would hope that he is? Not really as these stats change from season to season and only focus on one particular aspect of the game – the ability to convert shots into goals but it is interesting to see how, in this case, a goal is not just a goal and a shot is not just a shot.