What Is a Heatmap?

A heatmap is a visual representation of where a player or team spent their time on the pitch during a match. The concept is simple: every time a player touches the ball, makes a run, or occupies a particular zone, that action gets recorded as a data point. The pitch is divided into dozens of small zones, and every action in each zone adds to its colour intensity. The more activity in an area, the warmer the colour becomes. Bright red or orange means heavy presence. Cool blue or purple means barely any involvement.

The data behind these maps comes from two main sources. Event data records every time a player actively does something with the ball, a pass, a shot, a dribble, a tackle. Tracking data goes even further, recording the position of every player on the pitch up to 25 times per second, whether they have the ball or not. The combination of both produces heatmaps that capture not just where a player touched the ball but where they moved, where they pressed, where they held their position.

A single glance at a heatmap can tell you things that 90 minutes of watching a match might not make obvious. A striker whose heatmap is concentrated in a narrow central channel is a penalty box poacher. A striker whose heatmap spreads across the full width of the attacking third is a pressing forward who drops deep and wide. Same position, completely different job.

What Is a Pass Map?

While a heatmap shows where, a pass map shows how. Specifically, it shows the connections between players: who passes to whom, how often, and from which areas of the pitch. In a typical pass map, each player is represented by a dot positioned roughly where they spend most of their time on the pitch. Lines connect the dots to show passing relationships, and the thickness of each line represents how many passes were made between those two players. The thicker the line, the stronger the connection.

What makes pass maps so revealing is that they expose the skeleton of a team’s playing style. A team built around short passes through the middle will produce a dense network of thick lines clustered in central areas. A team that plays direct, using long passes to bypass the midfield, will produce a sparse network with thin lines in the centre and thick connections skipping straight from defence to attack. You do not need to know anything about tactics to look at two pass maps side by side and immediately understand which team controls the ball and which team gives it away quickly.

Pass maps also reveal where a team’s most important relationships are. If one particular line is dramatically thicker than all the others, that connection is the beating heart of the team’s build-up play. Remove one of those players through injury or suspension and the whole network changes shape, often in ways that explain a sudden dip in form.

Manchester City: The Most Recognisable Pass Map in Football

No team in the world produces more recognisable pass maps than Pep Guardiola’s Manchester City. Their network is immediately distinctive: a dense web of connections spread evenly across all areas of the pitch, with no single dominant relationship and no obvious weak link. Every player connects with almost every other player with roughly equal frequency. The lines are thick throughout.

This reflects Guardiola’s positional play philosophy, where every player must be comfortable receiving the ball in any situation and passing it to multiple options. The heatmaps that accompany City’s pass maps show players occupying precise zones with extraordinary discipline, the full-backs high and wide, the false nine dropping deep, the number eights arriving late into the box from midfield positions. The visual effect is of a machine: every part moving in coordination, no loose threads anywhere.

When City have an off day, you can often see it in the pass map before you see it in the scoreline. The network becomes lopsided, connections on one side of the pitch become thinner, the usual passing triangles break down. The data captures the messiness before the result confirms it.

Liverpool and the Full-Back Revolution

One of the most fascinating things pass maps revealed over recent years was the transformation of the full-back position. Liverpool under Jurgen Klopp, and now under Arne Slot, produce pass maps where the full-backs are among the most connected players on the pitch, often more so than the central midfielders.

Trent Alexander-Arnold’s pass maps in particular became famous in football analytics circles. His lines ran not just sideways to centre-backs or forward to wingers, but diagonally across the pitch, switching play with pinpoint precision. His heatmap showed him spending time in positions no right-back had traditionally occupied, sometimes almost in central midfield during the build-up phase. The numbers behind the visual confirmed what coaches and analysts were seeing: Alexander-Arnold was not a defender who occasionally attacked. He was a playmaker who occasionally defended.

This is exactly the kind of insight that pass maps deliver. Not just confirming what you already thought you saw, but revealing patterns that are invisible to the naked eye across ninety minutes of football.

Heatmaps That Told a Story: The Famous Examples

Some heatmaps have become famous in their own right because of what they revealed about a particular match or player. After Barcelona lost 8-2 to Bayern Munich in the 2020 Champions League quarter-final, Luis Suarez’s heatmap went viral. The striker had barely touched the ball outside the centre circle, spending most of the match returning for kickoffs after Bayern scored. The map was a perfect visual summary of a humiliation.

A similarly memorable case involved Andre-Frank Zambo Anguissa during a particularly dominant Napoli performance. His heatmap covered almost the entire centre of the pitch, from his own penalty area to the opposition’s, showing the relentless energy of a midfielder who never stopped running. The visual made the case for his performance more powerfully than any statistic alone could have.

Goalkeepers produce some of the most unusual heatmaps. A keeper who plays out from the back, sweeping behind a high defensive line, will have a map that extends well beyond the penalty area. A traditional keeper who stays on the line will have a map concentrated in a tiny box. Two players, same position, completely different profiles.

What Heatmaps Cannot Tell You

For all their power, heatmaps and pass maps have limits that are worth understanding. A heatmap shows where a player was, but not why they were there. A striker with a low-activity heatmap might have been tightly marked and effectively neutralised, or they might have been lazy and disinterested. The visual looks the same in both cases. Context matters, and context requires watching the match.

Pass maps show connections but not quality. A team can have a dense, impressive-looking network of passing connections and still be playing sideways and backwards for ninety minutes without creating a single chance. The map shows volume, not danger. That is why analysts almost always use pass maps alongside other metrics like expected goals, progressive passes, and chance creation to build a complete picture.

There is also the question of opposition influence. A team’s pass map against a deep defensive block looks completely different from their map against a high-pressing opponent. Comparing two pass maps without knowing the context of each match can lead to misleading conclusions. The best analysts always ask: what was the other team doing?

Where to Find These Maps Yourself

The good news is that heatmaps and pass maps are now freely available for almost every professional match in the world. Sofascore and Fotmob both offer player heatmaps on their free apps and websites, updated within minutes of a match finishing. FBref provides detailed passing networks for teams in the major European leagues going back several seasons. Understat offers shot maps and positional data for the top five European leagues.

The next time you watch a match, pull up the heatmap for the player you are most interested in at half-time. Then look again at full-time and notice how it changed. Did the pressing winger suddenly stop covering ground in the second half? Did the holding midfielder stop receiving the ball from the centre-backs? The map will often explain exactly why the result turned out the way it did, and in a way that no amount of commentary can quite capture.

That is the real power of these visuals. Not just showing you data, but making the invisible visible.

The Problem xG Was Built to Solve

Before xG existed, the most common way to measure a team’s attacking performance was shots on target. It was a reasonable proxy. Teams with more shots on target generally won more matches. But it had an obvious flaw: it treated every shot as equal.

A shot from six yards out with the goalkeeper beaten is not the same as a shot from twenty-five yards into a crowd of defenders. Both count as one shot on target. Neither the stat sheet nor the post-match summary could tell you which team had actually created the better chances. xG was built specifically to solve this problem.

The underlying question is straightforward: given everything we know about a shot at the moment it is taken, how likely is it to result in a goal? The answer is expressed as a probability between 0 and 1. A shot assigned 0.10 xG would be expected to be scored once in every ten attempts, historically, from similar positions under similar conditions.

Where the Metric Came From

The history of expected goals is more complicated than most accounts suggest. The term “expected goals” appeared as early as 1993, in an academic paper by Vic Barnett and Sarah Hilditch examining how artificial pitches affected goal-scoring rates in English football. Their model was rudimentary and focused on broader game patterns rather than individual shots.

A more direct ancestor was a 1997 paper by Richard Pollard and Charles Reep titled “Measuring the Effectiveness of Playing Strategies at Soccer,” which calculated weighted shot values based on distance from goal, angle, whether the shot was headed, and defensive pressure. This was xG in structure if not yet in name.

The metric entered football’s mainstream in April 2012, when Sam Green, an analyst at OptaPro, published a blog post asking how to quantify which areas of the pitch were most likely to produce goals. Green proposed a model that calculated the probability of each shot being scored. He gave each shot a value he called an “expected goal.” The abbreviation xG followed.

“Green does not regard himself as the inventor of Expected Goals. But he was in the right place at the right time to give football what would turn out to be its breakthrough metric.” — Rory Smith, writing on the history of football analytics

The metric spread through analyst blogs and Twitter in the years that followed. Its moment of mainstream arrival came in November 2017, when Arsene Wenger discussed an Arsenal loss to Manchester City using xG figures. He was among the first high-profile managers to reference the stat publicly. By 2017, Opta had made it an official metric. Today it appears on Match of the Day, Sky Sports, and virtually every major broadcast.

How xG Models Actually Work

There is no single universal xG model. Different data companies produce different xG figures for the same shots. Opta’s model, one of the most widely used, is built on close to one million shots from forty competitions spanning 2018 to 2022. It uses a machine learning method called XGBoost to calculate shot probabilities from more than twenty variables recorded at the moment of the shot.

The model does not measure what happens after the shot. It does not know whether the goalkeeper dived the right way, whether the ball hit the post, or whether the striker’s technique was good or poor. It only uses pre-shot information to estimate the probability based on historical precedent from similar situations.

This is a crucial distinction. xG is a measure of chance quality, not finishing quality. A striker who consistently scores chances with 0.10 xG is either finishing exceptionally well or getting lucky. Over a large enough sample, the data can begin to distinguish between the two.

Reading xG Correctly: Three Real Examples

xG Over Time: Why Sample Size Matters

This is where most fans misuse the metric. A team’s xG figures from a single match tell you relatively little. Football is a low-scoring, high-variance sport. Small samples produce misleading results. The research suggests the following rough guidelines:

Research by analyst Ben Torvaney found that a ten-match rolling window is close to optimal for using xG to predict upcoming attacking and defensive performance. Below that window, noise drowns out the signal. Beyond a full season, actual goals become nearly as reliable as xG for assessing quality.

What xG Cannot Tell You

The metric has clear and honest limitations. Understanding them is just as important as understanding what the metric does well.

It ignores goalkeeper quality. A shot assigned 0.30 xG is estimated based on the average goalkeeper. An elite goalkeeper saving that chance routinely does not make the chance a lower xG opportunity. This is why some analysts use post-shot xG models that incorporate where in the goal the shot ended up, giving a better measure of goalkeeper performance than pre-shot xG alone.

It cannot measure off-ball movement. The run that creates space for a teammate does not appear in xG. The pressing trigger that forces a turnover in a dangerous position does not appear in xG. Much of what makes teams good is invisible to the metric.

Different providers produce different numbers. Opta, StatsBomb, and Understat all run different models with different input variables and training data. Their xG figures for the same shot are often not directly comparable. When reading xG statistics, it matters which model produced them.

It does not explain why things happen. A team with persistently low xG is not performing well in chance creation. xG tells you that. It does not tell you whether the problem is tactical, personnel-related, injury-driven, or systemic. Explanation requires video analysis alongside the data.

Teams that dominate territory get inflated xG. Paris Saint-Germain in Ligue 1 will always have a low opponents-xG figure partly because they spend so much time in the opposition half, creating more pressing opportunities regardless of actual pressing intent. Context is always required alongside the raw numbers.

xG and Goalkeepers: The Separate Problem

Expected goals has produced a useful secondary application in goalkeeper analysis. By comparing the xG of shots a goalkeeper faces against the goals they actually concede, analysts can measure whether a goalkeeper is saving shots they should save, conceding goals they should not, or performing broadly in line with expectations.

The metric Post-Shot xG (PSxG) goes further, incorporating information about where in the goal the shot ended up. A shot aimed at the top corner has a higher probability of scoring than a shot aimed at the centre of the goal from the same position. PSxG accounts for this, giving a better measure of shot quality that factors in the difficulty of the save.

In the 1993/94 Serie A season, AC Milan’s goalkeeper Sebastiano Rossi went 929 consecutive minutes without conceding a goal. Modern PSxG analysis applied retrospectively to that defensive record would almost certainly show he was facing low-quality chances, reflecting the defensive organisation in front of him. That is xG doing one of its most useful things: separating individual performance from the system supporting it.

Why xG Appears on Your TV and What That Means

The arrival of xG on broadcast television was not universally welcomed. Jeff Stelling, the Sky Sports presenter, made a pointed comment in 2017 after a match where a team with a better xG figure lost. His frustration was genuine and his point was not entirely wrong: in a single match, xG can feel disconnected from what the scoreboard says.

The critics have a point about single-match xG. Showing one match’s xG figure during a broadcast can mislead viewers into thinking the team with higher xG deserved to win. Whether they deserved to win is a philosophical question. Whether their chance quality was superior is what xG answers. Those are different questions.

Used correctly, broadcast xG adds something useful. It tells a viewer when a scoreline flatters or punishes a team relative to the quality of chances they created. Nottingham Forest beating Tottenham 2-1 with 0.48 xG to Tottenham’s 2.14 is genuinely informative. It does not mean Forest got lucky. It means they converted with exceptional efficiency and Tottenham did not. Over time, those two things tend to equalise. That is the core insight the metric offers.

xG in Club Operations: How Professional Teams Use It

The most sophisticated use of xG is not in broadcast graphics but inside clubs. Liverpool integrated xG into their analytical operations around 2012, under data scientist Ian Graham, who joined as Director of Research. The club used the metric to inform recruitment, tactical planning, and performance review. Their subsequent rise under Jurgen Klopp, while not solely data-driven, was built on a foundation that took analytical models seriously.

Modern clubs use xG as one input among many rather than a single answer. A scout identifying a striker will look at their goals minus their xG over multiple seasons. A striker who consistently scores more goals than their xG suggests is either a genuinely elite finisher or overperforming and likely to regress. A striker who consistently underperforms their xG may be a finishing problem, a system problem, or a statistical anomaly. The data narrows the field of investigation. It does not close it.

The Bottom Line

Expected goals is not a perfect metric. No metric is. It is, however, the most useful single number for evaluating attacking and defensive performance in football over a meaningful sample of matches. It measures something real: the quality of the chances a team creates and concedes, estimated from historical data on millions of similar situations.

Its limitations are worth knowing. A single match’s xG figure is noisy. The metric cannot see off-ball movement, pressing organisation, or goalkeeper positioning. Different providers produce different numbers. And it does not explain anything, only describe.

What it does well, it does better than almost anything else available. Over ten or more matches, a team’s xG figures tell you more about their underlying quality than the actual scorelines do. That is a significant thing to know about any metric. Most statistics in football cannot make that claim.