Home / News / Gameweek review

Gameweek review · Gameweek 4

FPL GW4 review: Groß tops the scores with 17 points

Expected Pointless · Thursday 17 September 2026, 01:42 UK

Their projections for the next six gameweeks as the model has them now. Click a player for his numbers.

Groß was the top scorer of FPL GW4 with 17 points: a goal, two assists and maximum bonus for Brighton. Our model had him on 4.3, which makes him both the week's best player and its biggest miss. The FPL average score was 69, and the highest score in the game was 151.

FPL GW4 review: Groß tops the scores with 17 points
PlayerPointsGoalsAssistsBonusOur forecast
Groß BHA MID171234.3
Schade BRE MID152033.8
Bogle LEE DEF151034.0
Raya ARS GKP140034.0
Davis IPS DEF141212.5
Belloumi HUL MID132022.6
Dunk BHA DEF121004.0
João Pedro CHE FWD121136.1
Emersonn IPS FWD121132.8
De Cuyper BHA DEF110125.8

FPL Gameweek 4 highest scores

Groß led the week with 17 points from 1 goal, 2 assists and 3 bonus points in 90 minutes. Schade and Bogle were next on 15 each: Schade scored twice for Brentford, and Bogle scored from the Leeds defence in 82 minutes. Raya and Davis followed on 14, Raya with maximum bonus and Davis with a goal and two assists from the Ipswich back line.

Brighton had three of the top ten. Alongside Groß, Dunk scored 12 and De Cuyper 11. Chelsea's João Pedro and Ipswich's Emersonn also made 12, Emersonn in only 62 minutes.

Where did our model get Gameweek 4 wrong?

The biggest upside misses were players the model rated as ordinary options. Groß (forecast 4.3, scored 17), Davis (2.5, scored 14), Schade (3.8, scored 15), Bogle (4.0, scored 15) and Belloumi (2.6, scored 13) beat their forecasts by more than anyone else. Belloumi's 13 came in just 59 minutes.

The other side hurt more if you followed the model. Isak was forecast 6.8 and returned 2. Foden (forecast 4.3) and Reinildo (2.5) both finished on -2, Disasi (1.8) on -3, and Hall (4.2) blanked with 0.

Across the 307 forecast players who got minutes, the average miss was 2.08 points a player. Single gameweeks are noisy: the model is built to be right on average over a run of fixtures, and it is graded in public so you can judge whether it is.

How did the model's captain do?

Our model's captain was Haaland, and the armband returned 18 points. The full captain record, week by week, sits alongside the rest of the model's grading on the scorecard.

Us v the FPL Experts in Gameweek 4

Our team scored 83, beating the FPL Experts' 80. Our model's automatic XI made 71 and the FPL average was 69. Over the season, our team has beaten the FPL Experts in three of four gameweeks, but is 7 points behind them on total and 33 points above the FPL average.

How we worked this out

Points, goals, assists, bonus and minutes are FPL's own. Our forecast is the last projection the model published before the Gameweek 4 deadline, and the average miss covers every forecast player who got minutes. Every week is graded on the scorecard, the full detail is in the Gameweek 4 archive, and the season-long head-to-head is on us v the FPL Experts.

How this is written: an AI model drafts it from our projections and FPL's own data, and it is published only after an automatic check that every number, player and link in it comes from that data. The projections are graded in public on the scorecard.