The term expected goals (xG) gets thrown around a lot these days and people tend to take it as gospel truth, but should its value be in question? Before looking further into this matter, we need to know exactly what xG is and just as importantly – what it isn’t.
A single expected goal value represents the statistical chance that any given goal scoring chance will go in. It’s based on goal statistics from other chances from similar distances and angles. However, it does not take into account individual players’ abilities and preferences. For example, if a player like Juan Mata were to receive mostly high crosses he would likely have much fewer goals than predicted by xG.
In the case of keepers, in particular, xG can give us an idea of not just the number of saves made but the quality of saves made. In the 16/17 Arsenal vs Manchester United game, the Red Devils won 3-1. The expected goals of that game predicted 2-4 to Arsenal but it did not take into account an on form De Gea. De Gea made 14 saves that night including an insane double save where he saved a shot from Arsenal’s Lacazette and Sanchez’s ensuing follow up. This is where xG lets us down, De Gea is widely considered the best shot stopper in the world, meaning that he will always be likely to beat his xG in terms of a number of goals conceded and indeed last season De Gea conceded 16 goals less than expected. But xG failed to take into account the form and quality of De Gea.
This is why it is so important to take the context of the club, competition or environment into consideration when making predictions using xG. Another example of context that xG fails to take into account is Burnley’s defensive strategy. Their strategy allows for many incoming shots down the middle which typically possesses a high expected goals value, but their defenders position themselves in such a way that shots are funnelled straight down the middle where their keeper is positioned. This means they heavily outperform the expected goals against them due to their low block which operates contrary to other low block defences which typically enforce a policy of forcing players wide.
Expected goals also can’t predict a player’s mental state. Over the 17/18 season, Mohammed Salah had an xG90 value of 0,77 and achieved 0,97 goals per 90. This season he has a similar xG90 of 0,66 but because of mental pressure has only scored 0,47 goals per 90.
This is not to say xG does not deserve its popularity with pundits and statisticians alike. It can often give you a much better understanding of a match than just reading the scoreline would. For example, reading that Liverpool drew 0-0 with Manchester City could make one believe that it was an even game, but the xG paints a clearer picture. Liverpool had had their lowest xG of the season of 0.43 as City set up to nullify Liverpool’s attack. This also affected City, who also registered their lowest xG of the season with only 1.07 xG. But if it had not been for Mahrez’s fluffed penalty attempt, City would have walked away deserved winners.
By indicating a team’s quality and form, expected goals, can be used to make predictions about a team’s future form. Manchester City had the best xG over the 16/17 season and then walked the league the following season. Last season Manchester United’s xG suggest that they should have finished 6th, and currently they are tied in points with 6th place Bournemouth. Liverpool’s xG last season suggested they should be 2nd, and they too now have as many points as 2nd place, Chelsea.
Using expected goals is clearly a powerful tool when used properly; when keeping context in mind and understanding what it doesn’t tell you. Many football enthusiasts use this to make money from predicting football results. Players will usually first try a minimum deposit casino to get used to online gambling in a low-risk environment. Once comfortable and geared with the right stats websites and analytic skills they go on to betting on real football matches.
Nobody knows for sure what is going to happen in the world of football. In a sport where a single chance can turn an entire game on its head, no stat will ever be able to truly predict results. But xG is one of the best metrics to do so currently, it’s just important to interpret such metrics with first-hand insight from watching games.
