What are expected goals?

Melbourne Victory may have finished the 2018/19 Season empty-handed in terms of silverware, but in one stat they were the Hyundai A-League's runaway leaders.

Victory's Swedish forward Ola Toivonen's efficiency at making the most of the goalscoring opportunities that came his way made him the standout leader in a league table of players who exceeded their 'expected goals' tally for the campaign. 

While Golden Boot winner Roy Krishna netted 19 times compared to his expected goals total of 18.16 (an over-achievement of +0.84), Toivonen's 15 goals saw him over-perform by a whopping 7.14 above the tally of 7.86 suggested by the stats.    

Other players to defy the stats and score at a significantly higher rate than expected include Melbourne City's Ritchie de Laet (+4.48), Krishna's Wellington team-mate David Williams (+3.83), new Sydney FC signing Kosta Barbarouses (+3.53) and on-loan Central Coast Mariner Aiden O'Neill (+3.05).  

But what is an expected goal? And why does it matter? 

In an attempt to answer the pressing question of whether a goalscoring chance was really as good as we think it was, our good friends at Opta Data have come up with a computer algorithm to calculate 'expected goals'. 

Click below to #GetOnside for the FIFA Women's World Cup 2023™ in Australia

So, what are Expected Goals?

The phrase has entered the footballing mainstream in recent years, and its purpose is to assign a value (known as 'xG') to the quality of a goalscoring chance.

It does this by analysing data from thousands of shots and headers, and calculates the likelihood of a chance being converted, based on a range of factors including:

• Distance – how far away from goal the attempt is
• Angle – from where on the pitch the attempt is made
• Type of chance – was the shot taken from the ground, or was it a header or a volley?
• Type of assist – e.g. a long ball, cross, cutback
• Situation – is the player through one-on-one or is he/she being marked?

All of this information is put into the computer to create a percentage chance of a particular effort finding the net.

The higher the xG – with 1 being the maximum – the higher the likelihood that the chance will be taken.

For example, if a rebound falls to the feet of Sydney FC striker Adam Le Fondre inside the six-yard box and the chance is rated 0.7xG, you would expect it to be scored 70% of the time.

Which players outperformed their expected goals in 2018/19?

What else is xG used for?

As well as giving us an idea of how individual players are performing, expected goals can also be used to judge a team’s performance.

Take Adelaide United in 2015/16. The Reds did not win any of their first eight games in the Hyundai A-League, but their expected-goal difference throughout that run (scored vs conceded) was consistently higher than their actual goal difference.

The stats strongly suggested that the Reds were not playing as poorly as their results and standing on the ladder indicated.

And those stats, as it turned out, were pointing firmly in the right direction.

Starting with a 1-0 win over Perth Glory in early December, Adelaide’s fortunes duly changed and the team scored above their expected rate in all but one of their final 14 matches of the regular season.

Adelaide United's 2015/16 season: Goals (blue) vs Expected Goals (red) per game

Buoyed by that goalscoring efficiency, the Reds lost just once in their final 19 regular season games and went on to take out the 2015/16 Premiership and Championship double.

Evaluating Adelaide's xG showed that, behind a disappointing run of results, their underlying performance remained strong.