Explore the goals ↘︎Same goal.
Different odds.
A tap-in and a thirty-yard strike count the same on the scoreboard. This analysis covers all 362 goals from the Premier League’s top 30 scorers to see whose goals came from the least likely chances.
Morgan Rogers’ ten non-penalty goals came from the lowest-average-xG chances in the selected top 30.
Dominic Calvert-Lewin’s ten non-penalty goals came from the highest-average-xG chances in the group.
Richarlison at Arsenal: a goal from a chance rated at about 0.8% by Understat. The lowest individual xG in this cohort.
“Harder” means lower xG here. An xG of 0.10 means a model-estimated 10% scoring chance. Averaging only the shots that went in describes the goals a player scored. It does not, on its own, identify the better finisher.
One dot per goal. The diamond is the player’s average. Further left means a lower-probability chance. Select a player to inspect every goal.
30 players · 312 non-penalty goals · lowest mean xG first
xG describes chance quality in Understat’s model. This chart contains scored goals only. It is not a ranking of finishing ability. The top-30 selection breaks ties on fewer minutes; four further eight-goal scorers are available in the expanded view.
Aston Villa
10 season goals · 10 non-penalty · 0 penalties
Attacking right →︎ · Larger dots mean higher xG. Select a dot to find its goal below.
The 10 goals shown average 0.144 xG. Penalties are excluded; other set pieces and headers are included.
Across all non-penalty attempts, Morgan Rogers scored 10 goals from 7.15 xG: +2.85 against expectation.
The finishing comparison always uses all non-penalty shots, regardless of the goal filter. Neither figure alone establishes a player’s underlying ability.
| Date / fixture | Minute | xG | Situation / body part | Source |
|---|---|---|---|---|
| 19 Oct 2025Tottenham 1–2 Aston Villa | 36′ | 0.02528 | Open PlayRight Foot | Shot map ↗︎ |
| 23 Nov 2025Leeds 1–2 Aston Villa | 47′ | 0.08064 | Open PlayLeft Foot | Shot map ↗︎ |
| 23 Nov 2025Leeds 1–2 Aston Villa | 74′ | 0.10566 | Direct FreekickRight Foot | Shot map ↗︎ |
| 14 Dec 2025West Ham 2–3 Aston Villa | 49′ | 0.41485 | Open PlayRight Foot | Shot map ↗︎ |
| 14 Dec 2025West Ham 2–3 Aston Villa | 78′ | 0.01672 | Open PlayRight Foot | Shot map ↗︎ |
| 21 Dec 2025Aston Villa 2–1 Manchester United | 44′ | 0.03192 | Open PlayRight Foot | Shot map ↗︎ |
| 21 Dec 2025Aston Villa 2–1 Manchester United | 56′ | 0.06030 | Open PlayRight Foot | Shot map ↗︎ |
| 7 Feb 2026Bournemouth 1–1 Aston Villa | 21′ | 0.24484 | Open PlayRight Foot | Shot map ↗︎ |
| 19 Apr 2026Aston Villa 4–3 Sunderland | 46′ | 0.42591 | Open PlayLeft Foot | Shot map ↗︎ |
| 15 May 2026Aston Villa 4–2 Liverpool | 41′ | 0.03555 | From CornerRight Foot | Shot map ↗︎ |
Morgan Rogers averaged 0.144 xG across ten non-penalty goals. Harry Wilson averaged 0.151 across ten. Both scored six from below 0.10 xG. These are small samples, and the narrow gap is not evidence of a meaningful difference in ability.
Inspect Rogers’ goals ↗︎Across all ten goals, Cole Palmer averaged 0.497 xG. Remove his five penalties and it falls to 0.232. That is why the main ranking separates penalties—and why the explorer lets you put them back.
Inspect Palmer’s goals ↗︎Morgan Gibbs-White scored eight of his 14 non-penalty goals from chances rated below 0.10 xG. That is the most low-xG goals in the selected top 30, even though his overall average was not the lowest.
Inspect Gibbs-White’s goals ↗︎Richarlison’s goal away to Arsenal on 23 November 2025 carried an xG of 0.00805. It was the least likely goal in this selected group according to Understat’s model—not a claim that it was the league’s hardest goal by every measure.
Inspect Richarlison’s goals ↗︎The ranking above deliberately looks only at goals. For finishing performance, the denominator changes: non-penalty goals are compared with expected goals from every non-penalty attempt, including the misses.
Morgan Rogers scored ten non-penalty goals from 7.15 NPxG, a difference of +2.85. Dominic Calvert-Lewin scored ten from 14.87 NPxG, a difference of −4.87. The explorer shows this second measure alongside each player’s goal difficulty.
Scoring from close range can reflect movement, anticipation and a team’s ability to create good chances. A player whose goals average higher xG has not automatically had an easier job overall.
Roles, shot selection, teammates, model limitations and luck all affect these numbers. A single season’s goals-minus-xG is a description of that season, not a settled estimate of finishing talent.
Understat’s completed 2025/26 Premier League season: 380 matches. Players are ordered by goals, then fewer league minutes, then name. The selected top 30 scored 362 goals, of which 312 were non-penalty goals and 50 were penalties.
The cutoff is eight goals. Noah Okafor and Thierno Barry enter the 30 on fewer minutes. Nick Woltemade, Kiernan Dewsbury-Hall, Jaidon Anthony and Kevin Schade also scored eight; switch to all 34 to include them. That expanded sample contains 394 goals. It is not the entire league.
Mean goal xG = sum of the xG values of the goals shown ÷ number of goals shown. It is not total season xG divided by goals. The default excludes penalties but includes headers, free kicks and other set pieces. Open-play filtering uses Understat’s OpenPlay label. Own goals are excluded.
The orange dots use an editorial threshold of xG below 0.10. Diamonds mark the arithmetic mean; the player panel also gives the median. Goal locations use Understat’s normalised coordinates, oriented towards the right-hand goal. Dot area increases with xG approximately; the table gives the exact values.
Every included player’s goals, non-penalty goals, shots, total xG and NPxG reconcile with the source season totals (xG tolerance: 0.002). Match IDs restrict the data to this league season. Understat’s raw minutes may differ from broadcast minute labels.
These are pre-shot model estimates, not measurements of the quality of the strike. Different providers can assign different values. Small scored-goal samples and the exclusion of misses limit what this ranking can say about ability.
All downloads retain the four additional tied players and identify the selected top 30. Full xG precision is retained in the data; the page rounds comparison values to three decimals and individual goals to five.