Expected Goals Explained: How Our AI Uses xG to Predict Matches

If you’ve spent any time around football analytics, you’ve almost certainly heard the term “expected goals” — or xG, as it’s more commonly known. But what exactly is xG? How is it calculated? And why does our AI model use it as one of its key predictors? In this deep dive, we break down everything you need to know about the metric that has revolutionised football analysis.

What Is Expected Goals (xG)?

Expected goals is a statistical metric that measures the quality of a goalscoring chance. Every shot in a football match is assigned a value between 0 and 1, representing the probability that that particular shot will result in a goal. A penalty, for example, has an xG value of around 0.76 — meaning historically, 76% of penalties are scored.

A shot from the halfway line, on the other hand, might have an xG value of 0.01 — just a 1% chance of going in. The higher the xG value, the better the chance.

By adding up the xG values of all the shots a team takes in a match, you get their total xG for that game. This gives you a much better picture of how well a team played than simply looking at the scoreline.

How Is xG Calculated?

xG models are built by analysing hundreds of thousands of historical shots and looking at the factors that determine whether a shot goes in or not. The exact variables differ from model to model, but most include:

• Shot location: How far from goal was the shot taken? What angle?
• Body part: Was it a header or a shot with the foot?
• Type of chance: Was it a one-on-one? A rebound? A direct free kick?
• Game state: Was the team already winning? Losing? Drawing?
• Defensive pressure: How many defenders were between the shooter and the goal?

Our AI model uses a more sophisticated version of xG that also factors in things like player position data, passing sequences leading up to the shot, and even the quality of the assist. The more data points you include, the more accurate the xG value — and the better our predictions become.

Why xG Matters

Football is a low-scoring sport. A single moment of brilliance (or a single mistake) can decide a game. That means the scoreline doesn’t always reflect which team played better. A team might dominate a match, create chance after chance, and still lose because their opponent scored with their only shot on target.

xG helps us see beyond the scoreline. By looking at the quality of chances each team created, we can get a more accurate picture of how the game actually went. And over the course of a season, teams’ xG totals tend to converge with their actual goals scored — meaning xG is a much better predictor of future performance than current results alone.

How We Use xG in Our Predictions

At FutebolAI, expected goals is one of the foundational metrics in our prediction model. We don’t just look at how many goals a team has scored — we look at their xG, xGA (expected goals against), and the difference between the two. This tells us whether a team is overperforming or underperforming relative to the chances they’re creating and conceding.

For example, if a team is top of the league but their xG difference is only +5 while another team is +15, our model will know that the second team is actually playing better and is likely to finish higher over the full season. We call this “regression to the mean” — over time, results tend to align with underlying performance.

We also use xG at the individual player level. By comparing a player’s actual goals to their expected goals, we can see whether they’re overperforming (in excellent form, or perhaps lucky) or underperforming (struggling for form, or unlucky). This helps us predict how players are likely to perform in future matches.

The Limitations of xG

It’s important to note that xG isn’t perfect. It’s a probabilistic metric — it tells you what should happen on average, not what will happen in any specific game. And there are certain aspects of football that xG struggles to capture, like individual brilliance, tactical genius, and the impact of momentum.

That’s why our model uses dozens of different metrics, not just xG. We combine expected goals with data on possession, pressing intensity, set piece quality, defensive organisation, and many other factors to produce the most accurate predictions possible.

Next Steps

Expected goals has transformed football analysis over the past decade, and it continues to evolve. As more data becomes available — especially player tracking data — xG models will only get more accurate. And as our models improve, so will our predictions.

To see xG in action and explore how our AI uses it to predict match outcomes, check out our prediction accuracy page or dive into the full predictions record.

Disclaimer: This article contains educational content about football analytics and expected goals methodology. FutebolAI’s AI predictions are generated using a proprietary machine learning model incorporating xG and other metrics. All predictions are for entertainment purposes only and do not constitute betting advice. Always gamble responsibly.