About & Methodology

SoccerAnalytics.net publishes under its own name, not a person’s. It runs numbers on public football data and explains what they actually mean, with the working shown — no editorial team, no sponsor, no agenda beyond getting the maths right. Where an article says “we”, it means the site.

Who runs the site

The site is owned by a publisher who also runs other websites; they are listed on how this site is made, which also explains how the articles are made: from public data, under the owner’s direction. No article describes anyone’s personal experience of the game.

No club, league, broadcaster or data vendor has any say in what is published here, and none pays for a conclusion. The site is paid for by display advertising.

Editor

C. B. Zakarian (a pen name) is the owner and editor of SoccerAnalytics.net. C. B. Zakarian holds a Ph.D. in Information Systems, with a focus on data science, from the University of North Carolina at Greensboro (2016), and is an assistant professor.

Since October 6, 2026, every new piece on SoccerAnalytics.net is read and approved by C. B. Zakarian before it is published. Pages published earlier are being reviewed in weekly batches, and each reviewed page shows the date of its review.

Why the site exists

Expected goals and pressing metrics had leaked into mainstream commentary years before anyone bothered to explain them properly, and most of what passed for explanation was either hand-waving or marketing. The explanations here show the formula, run the code, and admit the parts that are still guesswork. When a model turns out to be wrong — and models are wrong constantly, often in interesting ways — that gets written up rather than quietly defended.

How the numbers are made

One rule sits above everything else: every statistic in every article traces back to a real source, and where a figure comes from a data pull the site runs, the dataset it was computed from gets published on this site next to the piece. Historical facts, formulas and published results are stated and cited; anything computed here comes with its inputs. If a number cannot be sourced one of those two ways, it does not go in. You will not find an invented figure dressed up as a measured one on this site. When data is not available yet, the article says so and talks about method instead of pretending to have results.

The data comes from public sources:

• StatsBomb open data — free, deep event data (every pass, shot, carry and pressure, with xG and freeze-frames) across a good spread of competitions, including several World Cups and Champions League seasons. It does most of the heavy lifting here. Anything built from it carries the attribution they ask for: "Data: StatsBomb open data."
• Understat — season-long xG for the big European leagues, used when a piece needs full-league tables the open event data doesn't cover.
• ESPN’s public scoreboard and match-summary feeds — the 2026 World Cup results and the Premier League box scores behind the season pieces.
• FBref (Opta/StatsPerform numbers) — linked from articles as a reference.

The tools are deliberately dull and reproducible — Python, statsbombpy for the data, mplsoccer and matplotlib for the pitch maps, pandas for the crunching. Each data-driven piece closes with the recipe written out in plain language — which file, which rows, which arithmetic — so you can redo the calculation with whatever tools you already use. Charts and tables get captioned with their source and the date the data was pulled — say, "Data: StatsBomb open data, retrieved June 2026" — because football data gets revised and a number is only as good as the snapshot behind it.

Check the numbers yourself

Transparency here is concrete, not a slogan. The raw dataset behind every 2026 World Cup piece — ESPN's public scoreboard and match-summary feeds, collected into one JSON file — is served on this site at /data/wc2026_results.json. Those are the same bytes the articles are computed from, so any figure can be re-derived, or challenged, without asking anyone’s permission. The 2022 deep dives work the same way from StatsBomb's free event data: the shot file, the match file and the 2022 final's full event feed are all served at the data_layer/… paths the articles name. The longer data pieces then close with a “Reproduce it” section giving the exact file and the computation step by step — the working is on the page, not hidden inside a program you would have to take on trust.

One thing to be plain about, because a half-kept promise is worse than none: those “Reproduce it” sections name the chart-drawing program that runs at build time — charts/chart_wc2022_headers.py and the like — and those programs are not published. They are named so you know which piece of machinery produced which figure, not as files to download. What is published is the input and the method: the dataset at the path the article names, and the computation written out in words. Between them you can reproduce any number on this site without ever seeing the code, which is the part that actually matters.

The models aren't just described — they're runnable. The tools page has interactive calculators for xG → match outcome, Elo win expectancy and shootout odds; the World Cup 2026 simulator Monte-Carlos the full 48-team format from strengths you choose; and the shot-map explorer puts every shot of the 2022 World Cup on one filterable pitch. Before the site’s prediction model said anything about 2026, it was backtested on 2022 and the result published — including a section on what the model could not do.

What the site is skeptical of

Most of how these numbers get used deserves distrust. People quote a single match's xG as if one game tells you anything — it mostly doesn't; the samples are tiny and the variance is enormous. "Underlying numbers" gets wheeled out to defend whatever the speaker already believed. And there's a tendency to treat any model output as more real than the football it came from, when half the value of a model is watching where it breaks. The same skepticism applies to the work on this site first.

What you can expect (and what you can't)

Plain language, real worked examples, honest uncertainty, and the occasional dry joke. Historical facts, formulas and rule changes are simply stated. Anything dressed as a current or recent-season stat is pulled fresh and dated.

No betting tips. No picks, no tipster content, no sportsbook promotions anywhere on the site. Where odds or betting markets do come up, they are the object of the analysis, never a recommendation — probabilities here are analytical tools and nothing more.

Corrections

Spot an error — a mislabelled axis, a stale number, a shaky assumption — and tell us; it will be fixed and the change noted. Being corrected efficiently is most of what good analysis is. Reach the site via the contact page or at contact@socceranalytics.net.

Last updated 6 October 2026.