Coventry beat Sunderland 3-1 on Saturday with three goals from chances worth 0.69 xG in total. Sunderland created 1.82 and lost. Before this match, Coventry’s typical return was no goals at all: their median in gameweeks 1 to 8 was zero. So this was the odd one of gameweek 9. Look at the whole season, though, and it seems less like luck than the moment the luck evened out.
Paid back
After four games Coventry were 4.19 goals behind their chances. After nine they are 0.70 ahead.
Source: Fantasy Premier League public data, 2026-27 gameweeks 1 to 9 (confirmed). Team xG is the highest xG conceded among the opponent's players.© simpitch.co.uk
How do you score three from 0.69 xG?
Each goal came from a chance FPL rated at under one in five:
- Sakamoto, 17’. His whole match was worth 0.07 xG. Trai Hume’s clearance went straight to him and he finished first time into the bottom corner. It was his first goal of the season.
- Awoniyi, 36’. His match was worth 0.18 xG. He took the ball off Luke O’Nien about 40 yards out, ran into the box and finished past Robin Roefs.
- Rudoni, 52’. His match was worth 0.19 xG. His shot from the edge of the box took a heavy deflection off Omar Alderete. Victor Torp, one of Frank Lampard’s four changes, was credited with the assist.
xG prices the shot, not the mistake that led to it or the deflection after it. Two of the three goals came from Sunderland errors, and the third changed direction on the way in. None of those things show up in a chance’s value. Sunderland head coach Régis Le Bris pointed to two unforced errors and said his side’s inconsistency was not acceptable at this level.
At the other end, Sunderland’s biggest chance was Enzo Le Fée’s 67th-minute penalty, after Ethan Pinnock was penalised for handball. Le Fée’s 0.85 xG is mostly that penalty. FPL doesn’t split it out, so Sunderland’s open-play total is unverified. Sky Sports’ figures, which do take the penalty out, have the open play almost level: Sunderland 0.80, Coventry 0.71. That comes from a different xG model, so it isn’t directly comparable with FPL’s.
The numbers
- 3.3%. If you treat 0.69 xG as a simple Poisson average, that is the chance of a team scoring three or more.
- 12 of 363. Across the 2024-25 and 2025-26 Premier League seasons plus this season’s first nine gameweeks, that is how many matches under 0.75 xG ended with the team scoring three or more. That is also 3.3%, so the data agrees with the model. It happens about once in thirty such matches.
- Second this season. It is only the second time in 2026-27 that a team has scored three or more from under 1.0 xG. The other was Spurs, against Coventry, in gameweek 7.
- 13.5%. On these chances the simple model gives Coventry a 13.5% chance of winning, against 64% for Sunderland.
Replay Saturday
On these chances the simple model gives Coventry a 13.5% chance of winning, against 64% for Sunderland. Coventry win about 1 time in 7.
The match replayed 10,000 times from each side's xG. Bars: share of replays. Black ticks: the exact model. Grid: how often each score came up; Saturday's 3-1 is outlined in red.
FPL xG for Saturday's match.
Source: Fantasy Premier League public data, 2026-27 gameweek 9; Sky Sports for the non-penalty xG. Simple Poisson model, seeded so every reader sees the same replays.© simpitch.co.uk
Not luck, a correction
Coventry came up from the Championship last season, after 25 years away, so there is no Premier League season of theirs to compare this with. Their own season tells the story instead. Lampard said afterwards that the side had started well and been rewarded for it.
| Gameweeks | Goals | xG |
|---|---|---|
| 1 to 4 | 0 | 4.19 |
| 5 to 9 | 9 | 4.11 |
| Season | 9 | 8.30 |
They created about the same in both spells. The goals just arrived late. Over nine matches they are less than one goal ahead of their chances, which is about as normal as finishing gets.
In short: Saturday was lucky, but Coventry’s season isn’t.
Where 0.69 xG usually lands
12 of 363 (3.3%): about one in thirty.
Every Premier League team match under 0.75 xG, one dot each, by goals scored. Orange: three or more. Red: Coventry v Sunderland. Hover any dot for the match.
Source: Fantasy Premier League data, from the vaastav/Fantasy-Premier-League archive for 2024-25 and 2025-26 (MIT licence) and FPL public data for 2026-27 gameweeks 1 to 9.© simpitch.co.uk
What these numbers don’t say
FPL gives xG per player per match, not per shot. Each scorer’s figure is their whole match, so it is an upper limit for the goal itself. We can’t see where the shots were taken from.
Nine matches is a small sample. And a 3-1 win against the bottom side, without a win in seven, says as much about Sunderland’s mistakes as about Coventry. The win moved Coventry from 19th to 15th.
Alderete’s red card came in the final minute of stoppage time, so it played no part in the result.
Last week’s forecast, checked
Last week we expected about 0.37 xG for Nicolas Jackson against Fulham and gave him roughly a 31% chance of scoring. He had 0.33 xG, scored before half-time in a 1-1 draw and finished with 9 points, one more than FPL’s projection of 8.0. The chances came in almost exactly as expected, and the goal fell on the right side of a roughly one-in-three call.
Next week: Coventry at Everton
Coventry visit Everton on Friday 6 November (20:00).
- Coventry: about 0.92 xG. That is their average of 0.92 xG a match, unchanged because Everton concede exactly the league average (1.44 a match).
- Everton: about 1.45 xG. This starts from Everton’s average of 1.38, then adjusts for Coventry conceding 1.52.
That gives:
| Outcome | Chance |
|---|---|
| Everton win | 50% |
| Draw | 27% |
| Coventry win | 24% |
| Coventry score at least once | 60% |
| Coventry score three or more again | 7% |
Friday, simulated
The forecast gives Everton 50%, a draw 27% and Coventry 24%. Coventry score at least once 60% of the time, and three or more 7%.
Coventry at Everton replayed 10,000 times from the forecast xG: Coventry 0.92, Everton 1.45. Bars: share of replays. Black ticks: the saved forecast. The model adds no home advantage.
Source: Fantasy Premier League public data, 2026-27 gameweeks 1 to 9; SimPitch forecast. Simple Poisson model, seeded so every reader sees the same replays.© simpitch.co.uk
The model doesn’t add a home advantage, so if anything it flatters Coventry. Treat it as a rough guide. We’ll check it next week.
Three takeaways
- Three goals from under 0.75 xG happens in about one match in thirty.
- Coventry went four games without a goal from 4.19 xG, then scored nine from 4.11. Across the season they’re level with their chances.
- At Everton on Friday, expect fewer than one goal’s worth of chances and roughly a one-in-four chance of a win.
Sources and licences
- Fantasy Premier League public data: goals, xG, xGC, minutes, points, cards and fixtures for 2026-27 gameweeks 1 to 9 (confirmed), and FPL’s projected points. SimPitch is not affiliated with the Premier League, Coventry City, Sunderland, Everton or Aston Villa.
- Fantasy Premier League data for 2024-25 and 2025-26: every team match’s goals and xG, from the vaastav/Fantasy-Premier-League archive (MIT licence).
- Match reports: linked and paraphrased, not quoted:
- Flashscore and Flashscore UK: how the goals were scored, the red card and the table.
- Sky Sports match report: goal minutes, the non-penalty xG figures and the managers’ comments.
- Coventry Observer: Lampard’s changes and Torp’s assist.
- beIN Sports: Le Bris’s reaction.
- Sky Sports: Coventry’s promotion.
- Calculations:
- Team xG for a match is the highest xG conceded among the opponent’s players.
- Probabilities use a simple Poisson model on that xG.
- The Everton forecast scales each side’s xG per match by the other side’s xG conceded per match, relative to the league average.
- The interactive charts replay matches in your browser: 10,000 draws from the same Poisson model, with a fixed random seed. Each chart’s sentence and labels are written from the data by fixed rules. No AI service is used.
- Photo: Coventry Building Society Arena, 29 November 2025, by Asj1977 via Wikimedia Commons, dedicated to the public domain under CC0 1.0 (Creative Commons CC0 1.0 Universal Public Domain Dedication), resized.
- Charts, text and calculations: made by SimPitch from the data above. © simpitch.co.uk.
Corrections
No corrections yet.
SimPitch (2026). Coventry 3-1 Sunderland: small change, big payout. simpitch.co.uk, 4 November 2026. https://simpitch.co.uk/weekly-analysis/posts/odd-gw9-coventry/