xG is a probabilistic estimate, not a prediction of what will happen. It becomes a reliable measure of team performance over larger samples — typically 10 or more matches — but is noisy for single games.

What is the difference between xG and xGoT?

xG measures the quality of the chance based on the shot's location and context. xGoT (or post-shot xG) additionally measures where the shot ended up in the goal frame, capturing execution quality.

What is a good xG per 90 for a striker?

In the top five European leagues, an elite centre-forward typically posts 0.55–0.85 xG per 90 minutes. Erling Haaland has exceeded 1.0 in Premier League title-winning seasons — an exceptional number.

Is xG useful for a single match?

Only cautiously. Single-match xG is noisy, and a team can genuinely deserve to win despite lower xG if their chances came from higher-quality build-up. xG is most useful over rolling 10-match samples or full seasons.

Expected Goals (xG) Explained

Q: What does xG mean in football?

xG stands for Expected Goals — a metric that estimates the probability that a given shot will result in a goal, based on historical data about similar shots.

What Expected Goals measures

Expected Goals — usually shortened to xG — is a metric that estimates the probability that a given shot will result in a goal. A shot given an xG value of 0.20 would be scored, on average, 20% of the time by the average finisher. Add up every xG value in a match or season and you get an estimate of how many goals a team should have scored, given the chances they created.

xG replaced 'shots on target' as the default measure of chance quality because it accounts for what actually matters — where the shot was taken from, how it was set up, and how difficult it was to convert.

How xG models are built

An xG model is trained on hundreds of thousands of historical shots. For each shot, the model records inputs such as distance from goal, angle to goal, body part used (foot, header), type of assist (through-ball, cross, rebound), whether the shot is from a set piece, and whether it is a fast break. The model outputs a probability between 0 and 1.

Different providers — Opta, StatsBomb, Wyscout — use different inputs and different training data, so their xG values for the same shot will differ. The direction of travel is the same but the exact number is model-dependent.

What xG tells you

Over a large enough sample — usually 10 or more matches — a team's cumulative xG is a much better predictor of future performance than the goals they have actually scored. A side scoring far above its xG is likely to regress; a side scoring below its xG is likely to score more soon. This is the single most useful application: identifying teams and players whose surface-level results are unsustainable in either direction.

xG also tells you about process. A team creating 2.0 xG per match but losing on the day is doing something right that the scoreline hides.

What xG doesn't tell you

xG measures shot quality — but a team's biggest weakness might be not getting into shooting positions at all. It cannot judge context: a shot taken with the defender's boot in the striker's ankle carries the same base xG as a free shot. Elite finishers (Erling Haaland, Kylian Mbappé) do consistently outperform their xG over long samples, so 'expected' is not always the right descriptor for the best strikers.

Single-match xG is also noisy. Do not use it to describe one game; use it across a season.

Post-shot xG (xGoT)

Post-shot Expected Goals, or Expected Goals on Target (xGoT), refines the model by including where the shot ended up in the goal frame. A shot placed in the top corner has a much higher xGoT than a shot straight at the keeper, even if both came from the same position. Comparing xG (chance quality) with xGoT (execution quality) reveals whether a team is being denied by great goalkeeping or by poor finishing.

Common misuses

The most common error is judging a single match by xG and declaring the losing team should have won. Real matches are not average outcomes: a 1-0 loss with an xG differential of 2.1 to 0.5 might mean the losing team was unlucky, but might mean their chances were poor-quality shots from tired legs against a well-organised block.

A related mistake is treating an xG chart as the whole story. Football is a chain of decisions; xG measures the last link. Progression metrics, pressures, defensive actions and pass networks describe the earlier links.

How xG changed football coverage

Ten years ago, an xG chart in a match report was a novelty; today it is standard on nearly every broadcast. Coaches use xG in team meetings, recruitment departments use it to identify undervalued attackers, and betting markets have integrated it into pricing. The metric has genuinely changed how the game is analysed — and, more importantly, has raised the standard of football journalism by adding evidence to the tactical debate.

How to read xG responsibly

Use season-long or rolling 10-match xG as a real signal. Treat single-match xG as one data point among many. Compare a player's xG with their actual goals over 20+ matches, not five. When a headline claims a striker 'should have scored', ask: what was the xG value of that specific chance, and how many chances at that value does a good finisher miss in a season? Applied carefully, xG is one of the most powerful analytical tools football has produced in a generation.

A worked example

Imagine a team creates the following four chances in a match:

A close-range header from a corner: xG 0.35
A shot from just outside the box, one defender in the way: xG 0.08
A penalty: xG 0.79
A speculative 30-yard strike: xG 0.03

The team's total xG for the match is 1.25 — roughly 'they should have scored one goal, occasionally two'. If they score three, they have over-performed the model and would be expected to regress over the next matches. If they score zero, they were unlucky but their process was decent — a coach can point at the numbers to argue against a knee-jerk change.

xG providers compared

| Provider | Inputs used | Public availability | |---|---|---| | Opta / Stats Perform | ~13 inputs incl. defensive pressure | Broadcast overlays, subscription | | StatsBomb | ~30+ inputs incl. shot freeze frame | Free open data, commercial API | | Wyscout | ~10 inputs, opta-family model | Subscription | | Understat | Reduced input set | Free public site |

Different providers can give the same shot noticeably different xG values, which is why analysts avoid comparing xG numbers across sources. What matters is consistency: pick a provider and compare xG over time using that provider only.

How xG changed recruitment

Ten years ago, striker recruitment was dominated by goal totals. Today, elite clubs look first at xG per 90, xG-goals differential (how much a striker over- or under-performs the model), and shot map — where their chances come from. This let clubs identify players like Jamie Vardy (an early xG darling at Leicester), Erling Haaland (elite xG + elite xG over-performance), and undervalued forwards in modest leagues whose underlying numbers pointed to bigger careers.

Recruitment departments now include analysts whose whole job is to spot mismatches between headline goals and underlying xG — the arbitrage opportunities that produced signings like Ferran Torres, Ollie Watkins and Nicolas Jackson.

Common misconceptions

A single missed 'high xG chance' does not prove a striker is out of form. Every model expects even 0.5 xG chances to be missed half the time.
xG is not opinion. It is a probability from a model trained on hundreds of thousands of shots. Reasonable people can debate model inputs; the number itself is objective.
xG does not measure defensive quality directly. Metrics like xG conceded, xGA on the ball, and post-shot xGoT capture parts of defensive performance separately.