Understat xG football analytics — export

Understat xG data is expected-goals analytics for football, where xG assigns every shot a scoring probability from 0 to 1 based on factors like distance, angle and body part. Understat publishes player, team, match and shot-level xG for the top European leagues. Alongside raw xG it exposes xA (expected assists), npxG (non-penalty xG), xGChain and xGBuildup, letting you judge chance quality rather than just goals scored.

No. Understat does not publish an official public API. The site is server-rendered HTML and embeds its statistics as JSON inside <script> tags (parsed via JSON.parse). Community libraries like understatapi (Python) and worldfootballR (R) scrape that embedded JSON, and Apify actors do the same server-side. Because the page sends no cross-origin (CORS) header, a browser tool cannot fetch it directly — which is why this page is a query builder, not a live fetcher.

Understat covers six competitions: the English Premier League (EPL), Spanish La Liga, German Bundesliga, Italian Serie A, French Ligue 1 and the Russian Premier League (RFPL). Coverage runs from the 2014/15 season through the current campaign. In this builder you supply seasons by their start year — for example "2025" means the 2025/26 season and "2014" the 2014/15 season.

It builds a ready-to-run query and shows you a fixed example of the output shape — it does not return live Understat results in your browser. You choose a mode (such as league_players), leagues, seasons and a result cap, preview the JSON columns you'll get back, then run that query live on the backing Apify actor, which scrapes Understat server-side and returns the real rows for export to JSON or CSV.

The backing actor, understat-football-analytics by Omar Eldeeb (Apify handle constructive_calm), is free to start, then pay-as-you-go on a pay-per-event basis. The first batch of chargeable events per run is free, and there is a per-run cost cap to prevent accidental overages. Pricing scales with how many rows or shots you export. Check the actor's Store page for the current per-event rates before a large run.

A player-season row includes the full Understat model: xG, xA (expected assists), npxG (non-penalty xG), xGChain and xGBuildup, alongside conventional stats like minutes played, goals, assists, shots, key passes, appearances, position and yellow/red cards. xGChain and xGBuildup credit a player for the value of possessions they were involved in, with xGBuildup excluding the shot and key pass to isolate buildup contribution.

They use different models. FBref uses StatsBomb data, which factors in defensive pressure and the number of defenders between the shooter and goal — inputs Understat's model ignores. As a result Understat tends to be slightly more lenient, assigning marginally higher xG to the same shot. Differences are roughly 3% at the match level and 5–10% on individual shots, so stick to one source within a single analysis rather than mixing them.

Understat's support has previously stated its data is free for non-commercial use with attribution, though that stance can change and is not a license. Responsible access means using only publicly available pages, respecting rate limits and not bypassing any authentication. The backing actor throttles shot-level requests to be polite to Understat's servers. For commercial use, review Understat's terms and consider a licensed provider; this builder does not grant you any rights to the data.

Both are possession-value metrics in Understat's model. xGChain sums the xG of every possession a player was involved in — any pass, dribble or shot that ended in a chance counts toward their total. xGBuildup is the same idea but strips out the shot and the final key pass, isolating a player's contribution to building the move rather than finishing it. They help surface deep-lying playmakers and progressors who don't show up in goals-and-assists tables.

Yes — the backing actor supports a match_shots mode that returns shot-level data: normalized x/y pitch coordinates, the xG model value for each shot, the situation (open play, set piece, penalty), body part, result, and the score at the moment of the shot, plus derived distance and angle. This is the data behind Understat's shot maps. This page's default preview uses league_players, but you can switch modes in the builder.