Goals from Shots Calculator
How many goals should a team's shooting bring? Enter its shots and shots on target per game, and a straight line fitted to 20 Premiership seasons gives the usual answer.
Averages per game over a season. Shots include blocked ones; on target means the keeper had to save it or it went in. The average Premiership side recorded 12.6 shots and 4.3 on target a game in 2025/26.
Try an example
- Goals a game
- Over 38 games
In one match
- No goals
- One
- Two or more
Over a season the line is usually within about 7 goals of what a team really scores. Try "lots of shots, few on target": 3.4 more shots a game than the average side, all off target, add only about 9 goals a season. On target, they would add about 37.
How it works
A straight line through 250 team-seasons, 2000/01 to 2020/21, chosen so its misses are as small as possible:
$$\begin{aligned} \text{goals} = \;&{-0.380} \\ &+ 0.283 \times \text{on target} \\ &+ 0.073 \times \text{off target} \end{aligned}$$In plain football
- Every shot on target is worth about 0.28 goals: roughly one goal every three and a half on target, or about 11 goals over a season for one extra a game.
- Every shot off target, blocked or wide, is still worth about 0.07 goals, about one in 14. A miss isn't worthless: it says the team is getting forward.
- The chances of none, one, or two or more in a match assume goals arrive at a steady rate, as in the Poisson Match Predictor.
Where it goes wrong
- It counts shots, not chances. A tap-in and a 30-yard hopeful count the same, which is why some sides, Celtic and Rangers above all, beat the line season after season.
- How shots are recorded has changed. From 2020/21 sides were recorded with about two more shots a game without scoring more, so compare like with like.
- It's a season-long average. Over a handful of matches, luck swamps it.
How it was built and tested: How many shots is a goal worth? Linear regression
These figures come from a statistical model and are for analysis and education. Football remains uncertain and model predictions will frequently be wrong.