How can xG be used to identify good bets?

By Charlie Dear

xG glossary

Expected goals [xG] 

The number of goals a team or player would be expected to score based on the quality and quantity of shots taken. 

Expected points [xPts]

Based on the xG metric, expected points is a way of quantifying the probability of winning a game if it was replayed multiple times. This awards xPts based on the xG in the game, regardless of the actual final result. 

Expected goals per 90 [xG/90]

Expected goals per 90 minutes played by a specific player

Non-penalty expected goals [npxG]

Total expected goals, not including penalties.

Shots per 90 [Shots/90]

Shots per 90 minutes [Minimum 30 minutes played]

____________________________________________________________________________ 

‘Expected goals’ (xG) was first introduced in 2012 by Opta’s Sam Green. But it wasn’t fully embraced across the football landscape as a whole until about five years later. It is a metric designed to measure the quality of chances that take place in a specific match. Each shot is calculated to have a specific probability of converting into a goal.

How is xG calculated?

xG is calculated by utilising a series of different metrics to analyse how likely a goal will be on a scale of 0-1. For example, a shot with an xG value of 0.3 is generally expected to convert three in every 10 attempts. Whereas an xG rating of 1.0 would suggest all of those 10 shots would go in.

We’re going to delve into how the numbers relate to performance, including how the shots/90 correlate with xG as a metric. Plus, how this metric can also be used to look at the bigger picture, including using the xPts model to showcase the point differentiation in clubs’ performance. This can help bettors better understand any nuances and perhaps oversight from bookmakers, in an attempt to get the edge when betting on specific matches. 

What are non-penalty expected goal stats used for?

Non-penalty expected goal statistics aim to give a more accurate representation of a player's performance, comparing their actual goals scored per match with their expected goals contribution. Penalties, which have an xG of 0.76, will seriously increase the xG per 90 minutes and can blur the player’s performance compared to players who are not penalty-takers. This is why they are discounted.

The benefits of expected goals

Football is notoriously a low-scoring sport and therefore measuring performance, setting aside the goals, has been hard to analyse - with pure goals data often being misleading about what can be expected from a player or a team's overall performance over the longer term.

There are numerous occasions when a team that has created more chances ultimately loses the match. Basic goal data can often distort this performance and produce metrics that misinterpret the game state.

For example, total shots count an effort from the halfway line as the same as a shot from inside the six-yard box. At its most rudimentary level, expected goals calculate the percentage chance of a goal scored based on the player’s position when shooting. If they are going through a rough patch of form in front of goal but still getting those ‘goal-scoring’ opportunities, then you could ascertain that their ‘luck’ will probably change.

These are some of the most common metrics that expected goals take into consideration:

• Distance to goal
• Angle to goal 
• One-on-one with the goalkeeper
• Body part [e.g. foot or head]
• Type of assist [e.g. through ball, cross, cut-back]
• Pattern of play [e.g. open play, counter-attack, corner, throw-in, free-kick]

In recent Premier League seasons, it’s been noted that the number of shots from outside the area has reduced dramatically compared to that of 10-15 years ago.

When we examine the mean of the three most recent complete seasons for the stats of goals from outside the box it reads an average of 7.31 per club; 

Mean 

2023/2024:  7.15

2022/2023:  7.35

2021/2022:  7.45

But if we go back ten years it confirms that there were more long-range shots scored during this period of the Premier League with an average across three separate seasons of 8.9 per club.

Mean

2013/2014: 9.3
2012/2013: 8.2
2011/2012: 9.25

Could this be due to the game’s evolution and the increase of statistical analysis and performance data? The rise in significance of xG, other data sets in football and even new managers to the league has put extra attention on a team’s style of play. This may or may not have contributed to this fall in ‘pot-shots’ and a greater level of detail on structural positional play. 

Using expected goals as a predictor of future performance

From an early match in the 2024/2025 Premier League campaign, basic goals would indicate Fulham’s 1-1 draw with West Ham was evenly matched. However, when exploring the data it would suggest on most other occasions Fulham would have won the game by two or more goals.

Supported by their home fans, Fulham had 21 shots with an xG of 2.89 in comparison to West Ham’s 0.68 xG, but the game still resulted in both teams only scoring one goal. According to Understat, Expected Points [xPts] from this game would have had Fulham win the game, with a score of 2.70 to the Hammers’ 0.20. 

However, a better barometer could be throughout a domestic season or a specific tournament - such as the UEFA Champions League. FIRST used FBref’s statistics to analyse four players with the exact same number of shots during the 2023/2024 competition to highlight how xG can present a more detailed evaluation of a player’s big chance percentage.

Kane: 1065 mins

Foden: 684 mins

Dembele: 954 mins

Martinez: 531 mins

The player who scored the most goals in the competition alongside Harry Kane was Kylian Mbappe. He also netted eight goals, but from 1,080 minutes and 23 shots on target in his 12 games. Mbappe’s shots per 90 was an average of 4.0. 

This shows just how prolific Kane was in the UEFA Champions League last season. His eight goals from 13 shots on target were converted from a mere average of 2.80 shots per 90, significantly less than the three other players in our comparison chart. 

Paris Saint Germain’s Ousmane Dembele and Inter Milan striker Lautaro Martinez both scored two goals from 33 shots with 10 on target. However, it’s important to note the players’ minutes in the competition and their differing shots/90 record.

Dembele’s two goals came from 954 minutes, 423 more minutes than Martinez [531 mins]. Dembele’s low xG ranking was from only 3.31 shots/90, which highlights he isn’t getting himself into as many goalscoring positions as Martinez did [5.58 shots/90]. Whereas Martinez’s 5.1 xG rating suggests he was often in the correct position but was wasteful in front of goal. 

What are the limitations of expected goals?

One potential drawback when attempting to examine the data is that there isn’t a ‘one size fits all’ universal model for xG. Since the metric’s boom in popularity, many different analysis websites have attempted to find the edge in both performance and betting strategy and these sources have different parameters behind the figures they produce.

Elsewhere, while xG may be fair to present this data with impartiality and a good measure to compare and contrast over a specific period of time, it doesn’t account for individual players and the varying levels of their ability. For example, there could be a lack of player-specific information when comparing the quality of opportunity if Kane and, say, Southampton defender Jan Bednarek were in the same scoring position. The opportunity would be given the same xG rating, despite Kane’s prowess and goal record. 

How can expected goals help identify good bets?

This can be a successful practice when making individual prop bets like wagering on a first, last or anytime goalscorer. Taking a closer look at a player's npxG to see whether they are on a hot streak or struggling for goalscoring chances. If you’d like to take it a step deeper, a player or team’s xG rating can also be converted into a percentage –  0.2xG would be a 20% chance of scoring.

On this note, examining a team’s recent xG rate and forecasting their trajectory; for example, if a team are underperforming their xG. This is where the over/under goals market could come into play, with the xG data supplying you with the information to determine whether a team are likely to start converting their attacking dominance into goals and victories. This is when you might place a bet for a team to score over 2.5 goals, despite not doing so in recent matches. 

The Asian handicap market could also come into play here if you notice a trend that a team is conceding less than the average but is also carrying a great attacking threat. In this case, the team’s xG statistics might suggest a bet of -1 or -1.5 on the Asian handicap would be the value play. 

Ultimately, It’s a great tool to use to back up your own theories or to identify trends across the medium to long term. However, xG is best used to complement your understanding of a match alongside team news and analysis of form and fitness, rather than being the sole determining factor behind your bets.