A decade ago, arguing about football meant goals, assists and what you saw with your own eyes. Today a pundit can cite a striker’s expected goals, a midfielder’s progressive passes, and a team’s pressing intensity before the second half has started. The numbers have moved from club analysts’ laptops into podcasts, broadcasts and every group chat that argues about the weekend’s results.
They are genuinely useful, and they are also easy to misuse. Here is where football data comes from, what the main metrics actually measure and how an interested fan can start building their own analysis.
Where the data comes from
Most professional football data is collected in one of two ways. Event data records every action on the ball: each pass, shot, tackle and dribble, with its location and outcome. It is produced largely by trained analysts, sometimes from the press box at games, who log matches from video, supported increasingly by automated systems. Tracking data records the position of every player and the ball many times per second, using camera systems installed in stadiums or computer vision applied to broadcast footage.
Event data powers most of the statistics fans see. Tracking data is more detailed and much more expensive, and it remains mostly inside clubs, leagues and a handful of specialist providers. When a number appears on television, it has almost always come from a commercial data company rather than from the club itself.
The metrics worth understanding
- Expected goals, which estimates the probability that a shot becomes a goal based on its location, the angle, the type of assist, and other factors
- Expected assists, which applies the same idea to the pass that led to a shot
- Progressive passes and carries, which measure how often a player moves the ball significantly closer to the opponent’s goal
- Passes per defensive action, a common measure of how aggressively a team presses high up the pitch
- Shot-creating actions, which credit the two actions that lead directly to each shot
- Field tilt, the share of possession each team has in the final third, which captures territorial dominance
None of these is a verdict on a player. Expected goals, in particular, describes the quality of chances rather than the skill of the finisher, and a striker who outperforms it for one season may simply have been fortunate. Over several seasons, the pattern becomes more meaningful.
Free data for fans
A surprising amount of football data is freely available. Several providers publish open datasets covering complete competitions for research and education, public statistics websites display advanced metrics for the major leagues, and market value and transfer databases track squad and contract information across thousands of clubs.
The first rule is to prefer official downloads and documented open datasets wherever they exist. They are cleaner, they come with definitions of each metric, and using them respects the work and the terms of the people who produce them. Many statistics websites explicitly prohibit automated collection in their terms of use, and that deserves to be taken seriously.
Building your own dataset
Fans who want to go further, for example to track a league that the big providers cover poorly, often end up collecting public information themselves: results, line-ups, match reports and league tables from official competition websites. At that point the practical problems are technical. Websites limit how many requests a single connection can make, and a collection run that is quietly throttled produces gaps that look like missing matches rather than a failed script.
Analysts who collect data at any volume spread their requests across several addresses, and for public pages the usual choice is fast datacenter proxies, which keep collection quick and inexpensive. The courtesy rules matter more than the tooling: read each site’s terms and robots file, collect slowly, store every page once so you never request it twice, and never republish someone else’s data as your own.
Common mistakes when reading the numbers
- Treating one season, or a handful of matches, as a reliable sample
- Comparing players across leagues without allowing for the strength of the competition
- Ignoring team context, since a full back in a possession side and one in a counter attacking side produce very different numbers
- Using totals instead of per ninety minute figures, which rewards players who simply play more
- Forgetting that different providers define metrics differently, so numbers from two sources rarely match exactly
Where analytics still falls short
Event data captures what happens on the ball, but much of football happens off it. The run that drags a defender away, the positioning that prevents a pass from ever being attempted, and the communication that organises a back line are hard to measure from events alone. Tracking data helps, but even the most advanced models struggle with decision-making, leadership and the psychological side of the game.
That is why the best analysts use numbers to ask better questions rather than to settle arguments. A surprising statistic is an invitation to watch the footage again, not a replacement for it.
Sharing your analysis responsibly
If you publish charts or threads based on your own numbers, a few habits earn trust quickly. Name the source of every figure, state the period and the number of matches behind it, and show per-90-minute figures rather than raw totals. When a result surprises you, say so and explain what you checked before posting it.
Be generous with credit, too. Open datasets exist because providers and researchers chose to share them, and acknowledging that keeps the ecosystem healthy for everyone who enjoys this side of the game.
A practical way to start
Pick one team or one league you already watch closely. Track a small set of metrics across a season, compare them with what you see in matches and note where they agree and where they do not. After a few months, you will understand both the game and the numbers far better than by reading any leaderboard, and your arguments with friends will improve considerably.
FAQ
What does expected goals mean?
It is the estimated probability that a shot results in a goal, based on factors such as distance, angle and the type of chance. It measures chance quality, not finishing skill.
Where can fans find football data for free?
Open datasets published for research, public statistics websites, and transfer databases cover a great deal. Always check each source’s terms of use.
Is it legal to collect football statistics from websites?
It depends on the site’s terms and on how the data is used. Prefer official downloads and open datasets, and respect any prohibition on automated collection.
How many matches do you need for reliable stats?
More than most people think. Many metrics stabilise only over a full season or longer, especially for individual players.
COMMENTS