The Ultimate Guide to xG (Expected Goals) in Football

If you have watched or read about a football match recently, you have almost certainly encountered the acronym xG. Once dismissed as "nerd nonsense" by traditionalists, Expected Goals (xG) has transformed from an underground analytics tool into a mainstream metric used by Premier League managers, television pundits, sportsbooks, and casual fans alike.

But what exactly is it, where did it come from, and why does it matter? Here is everything you need to know.

What is xG in Football Goals?

At its core, xG measures the quality of a goalscoring percentage chance. It calculates the probability that a specific shot will result in a goal, based on historical data from hundreds of thousands of similar shots.

Every shot taken in a match is assigned an xG value between 0 and 1:

0.00 means it is mathematically impossible to score from that situation.

1.00 represents a guaranteed goal.

For example, a penalty kick has an xG value of 0.76 because, historically, roughly 76% of penalties are converted. On the other hand, a desperate shot from 30 yards out through a crowded penalty box might have an xG of 0.02, meaning the player has a 2% chance of finding the back of the net.

Variables That Shape xG

Advanced AI models (like those from Opta) analyze over 20 context factors in real time to calculate a shot's value:

Distance and Angle: How close is the player to the goal, and how wide is the angle?

Type of Assist: Was it a through ball, a cross, a rebound, or a header?

Defensive Pressure: How close are the defenders to the shooter?

Goalkeeper Position: Is the keeper well-positioned, or are they caught out of their net?

By adding up the xG of every shot a team takes during a match, you get a cumulative total that shows how many goals they should have scored based on the quality of their chances.

Who Invented xG and When Was It Introduced?

The modern data revolution did not happen overnight. The phrase "Expected Goals" was first coined in an academic paper by Vic Barnett and Sarah Hilditch in 1993, which investigated the impact of artificial pitches.

However, the foundation for the metric dates back much further:

The 1950s & 60s: A British military accountant and football obsessive named Charles Reep painstakingly hand-recorded data across 667 matches. He discovered it took roughly 10 shots to produce one goal. While his work led to the (often criticized) "long-ball" era of English football, it was the earliest attempt to map out high-probability shooting zones.

The 2000s: Researcher Richard Pollard published a paper in 2004 that outlined specific variables affecting a shot's success rate.

The 2010s (The Breakthrough): The modern, algorithm-driven iteration of xG was popularized by early analytics pioneers like Sam Green (who created an xG model for Opta in 2012) and blogger Michael Caley. It crossed over to the mainstream in 2017 when BBC’s Match of the Day began featuring xG in its standard post-match graphic layouts.

What is xG in Football Premier League?

In the English Premier League, xG is used to look past the surface-level scoreline to evaluate how well a team is actually playing.

A traditional scoreline tells you who won, but it doesn't always tell you who dominated. A team could win 1–0 due to a lucky deflection, despite being thoroughly outplayed. By assessing the Premier League through an xG table, analysts can identify which teams are genuinely elite creators and which ones are riding a wave of good fortune.

Overperforming vs. Underperforming xG

Overperforming xG: If a striker has an individual xG of 10 for the season but has scored 15 goals, they are "overperforming." This usually indicates world-class finishing capability (think Erling Haaland or Mohamed Salah).

Underperforming xG: If a team creates high-quality chances but fails to score, their xG will be much higher than their actual goals. This indicates great tactical build-up play, but poor finishing.

Highest xG in Football History (Premier League Era)

When a team completely overwhelms their opponent, the underlying numbers go off the charts. Because comprehensive xG data collection only became standardized in the 2010s, records are primarily tracked from that point forward.

Highest Single-Team Match xG

The record for the highest xG generated by a single team in a Premier League match belongs to Liverpool. On January 1, 2024, Liverpool put on an attacking masterclass against Newcastle United, racking up an astonishing 7.39 xG. Despite creating chances worth over seven goals, the match finished 4–2, illustrating how xG captures the sheer volume of missed opportunities.

Rank	Club	Match xG	Opponent	Date	Final Score
1	Liverpool	7.39	vs. Newcastle	Jan 1, 2024	4–2
2	Liverpool	6.28	@ Tottenham	Dec 22, 2024	6–3
3	Manchester City	5.87	vs. Watford	Sep 21, 2019	8–0
3	Liverpool	5.87	vs. Leeds United	Feb 23, 2022	6–0

Highest Individual Player Match xG

The highest individual xG performance in a single Premier League match belongs to Crystal Palace striker Jean-Philippe Mateta. On October 18, 2025, Mateta registered an individual 3.48 xG against Bournemouth, scoring a brilliant hat-trick.

What is xG in Football Prediction?

Because football is a low-scoring sport heavily influenced by luck and deflections, actual results can be deceptive. This makes xG an incredibly powerful tool for football prediction and sports betting.

If you want to predict who will win next week's match, looking at a team's xG over their last 5 to 10 games is far more predictive than looking strictly at their wins and losses.

The Regression Principle: Over time, a team's actual goals scored will almost always gravitate toward their expected goals (xG).

If a mid-table team has won three games in a row 1–0, but their xG in those matches was only 0.40 compared to their opponents' 2.10, they are heavily relying on luck or miraculous goalkeeping. Smart predictors know that this form is unsustainable and that the team is due for a loss, making them a prime target for value betting.

While both metrics are critical tools in modern football analytics, the fundamental difference between Expected Goals (xG) and Expected Goals on Target (xGOT) comes down to a timeline: xG is a pre-shot model, whereas xGOT is a post-shot model.

1. The Core Differences

To put it simply, xG evaluates the quality of the chance, while xGOT evaluates the quality of the execution.

xG (Expected Goals): Measures the probability that a shot will result in a goal at the exact millisecond before the ball leaves the player’s foot. It does not care where the shot actually travels. A tap-in from two yards out has a massive xG, even if the striker accidentally kicks it out of the stadium.

xGOT (Expected Goals on Target): Measures the probability of a goal after the ball has been struck, factoring in the shot's final trajectory and end location within the goal frame. If a shot misses the target completely or is blocked by a defender, its xGOT immediately drops to 0.00.

The Perspective Shift: xG is usually analyzed from the attacker’s perspective (how good are they at getting into dangerous positions?). xGOT is usually analyzed from the goalkeeper's perspective (how difficult was the shot to save?).

2. How They Are Calculated

Both models assign a value between 0.00 (impossible to score) and 1.00 (a guaranteed goal), utilizing historical data from hundreds of thousands of past shots to calculate probabilities. However, they ingest entirely different variables.

How xG is Calculated

The data model focuses entirely on the setup and the environmental factors around the shooter:

Distance and Angle: How far is the player from the net, and how wide is the shooting angle?

Type of Assist: Was the chance created via a through-ball, a cross, a defensive rebound, or a corner?

Body Part: Is the player shooting with their dominant foot, weaker foot, or heading the ball?

Defensive Pressure: How close are the nearest defenders, and is the goalkeeper set and in position?

How xGOT is Calculated

The xGOT model starts with the baseline xG of the chance, but then introduces post-shot execution data captured by computer vision tracking systems:

Goalmouth End Location: Exactly where did the ball cross the goal line? A shot placed perfectly in the absolute top corner receives a dramatically higher xGOT than a shot fired straight down the middle into the keeper's chest.

Ball Velocity and Trajectory: How fast was the ball traveling, and did it feature any intentional swerve or deflection?

Goalkeeper Freeze Position: Advanced models (like Opta) calculate the goalkeeper’s precise distance from the ball's final trajectory line at the moment of the strike.

3. Side-by-Side Application Example

Imagine an attacking midfielder picks up the ball at the edge of the 18-yard box.

Scenario	xG Value	xGOT Value	Analysis
The Setup	0.05	N/A	Historically, only 5% of shots are scored from this far out under defensive pressure.
Scenario A: The player scuffs the shot slowly along the grass, straight into the goalkeeper's arms.	0.05	0.02	The xG stays 0.05. However, because it was an incredibly weak, poorly placed shot on target, the xGOT drops, meaning any average keeper saves it 98% of the time.
Scenario B: The player strikes it cleanly, curling it with high velocity into the absolute top-left bin.	0.05	0.85	The xG remains 0.05 (it was still a tough chance initially). But because of world-class execution, it becomes an 0.85 xGOT—a shot so well-placed that it results in a goal 85% of the time, putting immense pressure on the keeper.
Scenario C: The player tries to blast it, but slices it entirely wide of the goalposts.	0.05	0.00	The initial xG chance is still recorded as 0.05 for team attacking stats. But because it didn't hit the target, it has an xGOT of 0, requiring zero intervention from the goalkeeper.

4. Why Analysts Use Both

By combining these two metrics, teams can unlock deeper hidden insights:

Shooting Goals Added (SGA): By subtracting a player's cumulative xG from their cumulative xGOT ($xGOT - xG$), coaches can isolate a player's raw finishing ability. If their xGOT is consistently much higher than their xG, it proves they are actively transforming poor, low-probability chances into lethal, high-quality finishes through elite shooting technique.

Goals Prevented (Goalkeeping): By subtracting a goalkeeper's actual conceded goals from the total xGOT they faced ($xGOT - \text{Goals Conceded}$), you can accurately judge a shot-stopper. It filters out whether a keeper keeps clean sheets simply because they have a great defense, or because they are actively pulling off miraculous saves against "un-savable" top-corner rocket shots.