Skip to content

Expected goals (xG)

also: xG · shot quality

The chance a shot is scored, from where it was taken and whether it was a header, fitted on the league's own history.

Formula
xG = P(goal | distance, angle, header) ; penalties = the league's conversion rate

Every located shot carries its spot on the pitch. We group a league's shots by distance from the goal (2 m buckets), angle off the goal's centre line (15 degree buckets) and header or foot, and take each group's goal rate, shrunk toward the rate at that distance when the group is small. A penalty takes the league's penalty conversion.

Summed over a team's shots, xG measures the chances it made; summed over the shots it faced, the chances it allowed. Next to goals it separates making chances from finishing them, which is why it predicts the next run of results better than the table does.

The common mistake

Ours is a location model. It does not see the defenders, the goalkeeper's position or the pass before the shot, so a tap-in after a cut-back and a crowded six-yard scramble from the same spot get the same value. Treat a one-match xG gap as noise; read it over ten matches or more.

The data behind it

Expected goals (xG) comes from Premier League shot location and xG data and Premier League match event data. The location, end point, outcome and body part of every Premier League shot charted since 2012-13, with our expected goals. REST or MCP.

Where to get it

Related

Expected goals (xG) from the API

500 free credits every month, no card. REST or MCP.

curl -X POST https://mcp.strikerapi.com/v1/get_shot_map \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_API_KEY" \
  -d '{
  "game_id": "740916"
}'

get_shot_map: every parameter and the full response.

Also from StrikerAPI: player props history by sportsbook · today's props board · every dataset

Get a free key