{"id":4077,"date":"2021-11-11T12:56:10","date_gmt":"2021-11-11T12:56:10","guid":{"rendered":"https:\/\/www.ukgamblingsites.com\/?page_id=4077"},"modified":"2023-03-07T15:59:54","modified_gmt":"2023-03-07T15:59:54","slug":"what-are-expected-goals-xg-how-are-they-calculated","status":"publish","type":"page","link":"https:\/\/www.ukgamblingsites.com\/sports-betting\/football\/what-are-expected-goals-xg-how-are-they-calculated\/","title":{"rendered":"What Are Expected Goals (xG) & How Are They Calculated?"},"content":{"rendered":"

\"xGDepending on where you stand, you might view expected goals (xG) as an insightful statistical measure or an overly scientific waste of time. The rather rapid rise of xG to mainstream football analysis has certainly divided opinion along the way.<\/p>\n

Whether you value the metric or not, it looks like xG is not going away anytime soon. As such in this article, we\u2019ll be explaining exactly what the expected goals statistic is and what it tells us.<\/p>\n

We shall also examine some of the limitations of xG models. Although there are some limitations, as it is by no means a flawless metric, some of its criticisms come from a limited understanding of what the data represent.<\/p>\n

For those that know exactly what the xG figures show, most would agree that it has definite value and it is a tool that can even aid your betting selections.<\/p>\n

Expected Goals – Why Is It Needed?<\/h2>\n

\"xG<\/p>\n

Before discussing how expected goals works, we should first ask why are we bothering with it at all? Are shots on target and off target not sufficient for giving us an indication of which side has been the more potent in attack over the course of 90 minutes? Measuring shots is often a good indicator of judging the \u2018best\u2019 team but it is by no means perfect. Let us highlight the limitations of relying solely on more traditional stats through a couple of examples.<\/p>\n

The first game we wanted to highlight was a Premier League contest between Tottenham and Leicester played on 19th July 2021. The stats are as follows.<\/p>\n\n\n\n\n\n\n\n\n
Stat<\/th>\nTottenham<\/th>\nLeicester<\/th>\n<\/tr>\n<\/thead>\n
Possession<\/td>\n29%<\/td>\n71%<\/td>\n<\/tr>\n
Shots<\/td>\n7<\/td>\n24<\/td>\n<\/tr>\n
Shots on Target<\/td>\n3<\/td>\n6<\/td>\n<\/tr>\n
Corners<\/td>\n4<\/td>\n13<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Judging by these stats alone, you would think Leicester won the game comfortably when in actual fact they lost 3-0. Spurs raced into this commanding lead after 40 minutes and then effectively set up a brick wall, forcing Leicester into many speculative efforts from outside the box or from tight angles. Eleven of the Foxes\u2019 attempts came from outside the box and this meant their xG only went as high as 1.20, an average of 0.05 per effort.<\/p>\n

Another example comes from a clash between Manchester City and Crystal Palace during the 2020\/21 season. The Citizens were the strong favourites for the fixture and a look at the stats suggests that the game went completely as expected.<\/p>\n\n\n\n\n\n\n\n\n\n
Stat<\/th>\nMan City<\/th>\nCrystal Palace<\/th>\n<\/tr>\n<\/thead>\n
Goals<\/td>\n4<\/td>\n0<\/td>\n<\/tr>\n
Possession<\/td>\n72%<\/td>\n28%<\/td>\n<\/tr>\n
Shots<\/td>\n13<\/td>\n2<\/td>\n<\/tr>\n
Shots on Target<\/td>\n6<\/td>\n0<\/td>\n<\/tr>\n
Corners<\/td>\n11<\/td>\n1<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Nothing untoward here but xG actually paints a distinctly different picture of this game. City\u2019s expected goal\u2019s totalled just 0.96 meaning they were outrageously clinical during this contest. Given the quality of chances they had it would have been more likely that they would have won the game just 1-0, or even been involved in a goalless draw.<\/p>\n

What Does xG Show?<\/h2>\n

\"GoalTraditionalists among you might say that the only number that matters is goals and in a sense this is true. Expected goals, is definitely a statistic many find interesting though as it helps give a better sense of how competitive a match was. It does this by awarding a score for every single shot taken during a game. The score awarded depends not on the quality of the shot itself but the quality of the opportunity.<\/p>\n

The lowest score issued is 0.01 and this could be, for example, an effort from 40 yards out from a tight angle that would very rarely trouble a goalkeeper. An absolute sitter on the other hand could give an xG of 0.8 or 0.9. In theory, the maximum score is 0.99 as 1 would imply a certain goal, something which does not exist. The numbers directly represent the likelihood of a certain chance being converted. If a striker has a shot with an xG of 0.4, this means that they would be expected to score it 40% of the time. It does not necessarily mean this one-off strike has a 40% chance of success however because the quality of the striker and goalkeeper also makes a difference.<\/p>\n

By totting up how presentable each shot was, this will produce a team score for each game. Overall, there is a good correlation between actual goals scored and the xG. Take a look at any week and you will see this is the case. For an illustration though, here is gameweek 11 of the 2021\/22 Premier League season.<\/p>\n

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Team<\/th>\nxG<\/th>\nActual Goals<\/th>\nActual Goals<\/th>\nxG<\/th>\nTeam<\/th>\n<\/tr>\n<\/thead>\n
Southampton<\/td>\n0.9<\/td>\n1<\/td>\n0<\/td>\n0.54<\/td>\nAston Villa<\/td>\n<\/tr>\n
Man United<\/td>\n0.76<\/td>\n0<\/td>\n2<\/td>\n2.04<\/td>\nMan City<\/td>\n<\/tr>\n
Brentford<\/td>\n2.79<\/td>\n1<\/td>\n2<\/td>\n1.16<\/td>\nNorwich<\/td>\n<\/tr>\n
Chelsea<\/td>\n3.01<\/td>\n1<\/td>\n1<\/td>\n0.67<\/td>\nBurnley<\/td>\n<\/tr>\n
Crystal Palace<\/td>\n2.65<\/td>\n2<\/td>\n0<\/td>\n0.25<\/td>\nWolves<\/td>\n<\/tr>\n
Brighton<\/td>\n1.43<\/td>\n1<\/td>\n1<\/td>\n0.23<\/td>\nNewcastle<\/td>\n<\/tr>\n
Arsenal<\/td>\n1.49<\/td>\n1<\/td>\n0<\/td>\n0.49<\/td>\nWatford<\/td>\n<\/tr>\n
Everton<\/td>\n0.60<\/td>\n0<\/td>\n0<\/td>\n0.80<\/td>\nSpurs<\/td>\n<\/tr>\n
Leeds<\/td>\n1.97<\/td>\n1<\/td>\n1<\/td>\n0.80<\/td>\nLeicester<\/td>\n<\/tr>\n
West Ham<\/td>\n1.25<\/td>\n3<\/td>\n2<\/td>\n1.46<\/td>\nLiverpool<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

From this information, we can see that most teams were within a goal of their xG with the exceptions of Brentford and Chelsea who were wasteful and West Ham who were helped out by an own goal. Chelsea fans in particular can feel a little aggrieved that their team did not bag all three points but they happened to find Nick Pope in inspired form.<\/p>\n

How Is xG Calculated?<\/h2>\n
\"xg
From Analytics Vidhya<\/a><\/figcaption><\/figure>\n

We know that xG represents how good a shooting opportunity is, but how exactly are the figures produced? Well, xG uses masses of data to find out, historically, how often players have scored from incredibly similar situations. Opta, for instance, uses 300,000 shots for their own xG model so it is not simply a case of someone awarding a subjective rating while watching the game. An xG model will typically take into consideration the assist type, the shot type (so header, volley, and so on), distance from the goal, angle to the goal and the positioning of defenders.<\/p>\n

It is important at this stage to note that there is not simply one \u2018xG\u2019 calculation, like with shots on target, as models do vary. The figures used in the article are provided by Understat<\/a> but there are several other supplies such as Opta, Infogol and StatsBomb. In the case of StatsBomb, they made a significant adjustment<\/a> to their xG model in 2020 as it began factoring in \u2018shot impact height\u2019 which is the height of the ball at the time of the shot.<\/p>\n

It is therefore possible to see xG figures that differ depending on the source. For most shots though, they tend to produce very similar numbers.<\/p>\n

Player xG<\/h2>\n

\"Player<\/p>\n

People will say that xG is flawed because it does not take into account the quality of the person taking the shot. Robert Lewandowski would produce the same xG for taking a penalty as would Manuel Neuer. To consider this to be a flaw though is a misunderstanding of what xG is there to show. Expected goals is designed to tell you the chances of a (statistically) average player scoring in that position. Through this, you can then see which players are especially clinical and which are extremely wasteful.<\/p>\n

In the 2020\/21 Premier League season the most clinical player was Heung-Min Son who scored 16 goals despite an xG of just 11.02. This was the fifth season in the row the Korean had outperformed his xG in the league, with a total overperformance of 18.13 goals. By being so consistently a better finisher than average gives us \u2018proof\u2019 that he is among the best finishers in the world.<\/p>\n

By contrast, Timo Werner ended the same season with a -7.43 under performance, making him the most wasteful player in the division. This was surprising to see given that he had been extremely accurate in front of goal with RB Leipzig the season prior. Other players seem to be consistently inaccurate when shooting such as Manchester City\u2019s Gabriel Jesus as the league performances below show.<\/p>\n\n\n\n\n\n\n\n\n\n
Season<\/th>\nExpected Goals<\/th>\nActual Goals<\/th>\n<\/tr>\n<\/thead>\n
2020\/21<\/td>\n9.65<\/td>\n9<\/td>\n<\/tr>\n
2019\/20<\/td>\n21.02<\/td>\n14<\/td>\n<\/tr>\n
2018\/19<\/td>\n12.62<\/td>\n7<\/td>\n<\/tr>\n
2017\/18<\/td>\n15.37<\/td>\n13<\/td>\n<\/tr>\n
2016\/17<\/td>\n7.24<\/td>\n7<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Only thanks to xG do we have clear, objective evidence that Jesus is a wasteful player when shooting. Of course, anyone who watched his regularly could tell you this also but it helps to have more concrete proof.<\/p>\n

Limitation of xG<\/h2>\n

While the inability to factor in player quality is not, we would argue, a flaw of xG, this does not mean the metric is perfect. One significant issue it has is the inability to fully take into account the context of the shot. For the vast majority of shots, this is not an issue but xG can sometimes significantly underplays how presentable a chance is, often when a keeper is not in their usual position.<\/p>\n

Take a look at the example below in which Jesse Lingard basically has an open goal tap in after tackling Kasper Schmeichel in the Leicester net. At the time Lingard shoots, it would take a huge error on his part not to direct the ball into an empty net.<\/p>\n