GW7 reflection 2026/27
Published Tue 20 Oct · version 2
The week in five lines
- 1 · Team goals26 goals against 29.7 expected.Ratings updated
- 2 · Positions and playing time22 players we had at 70%+ to start didn't.17 start chances moved
- 3 · Injuries, suspensions and congestionLiverpool lost Szoboszlai and Frimpong.Rotation measured
- 4 · Set piecesNo change to FPL's set-piece takers.No change
- 5 · CaptaincyHaaland scored 13; the underlying-form pick, Saka, 3; last week's top scorer, João Pedro, 2.Premium 112 v xGI form 28
What the model changed this week
Updated automatically
Method changes since the GW7 deadline
None.
Watching and considering
- B.Fernandes at right wing: one match so far
- Palmer at attacking midfield: one match so far
- Mbeumo at centre-forward: one match so far
- Lower start chances after a European or cup midweek (not modelled yet; measured each week in this section)
- A source for who takes each penalty
1 of 5
Team goals
26 goals against 29.7 expected. Chances were down too (xG 27.1), so the shortfall was in chances created, not just finishing.
Expected v chances v goals, by team
| Team | Chart | Goals v exp | xG v exp |
|---|---|---|---|
| Brighton v CRY | +3.1 | +0.4 | |
| Ipswich @ MCI | +1.2 | +1.1 | |
| Tottenham v COV | +1.0 | −1.2 | |
| Bournemouth v SUN | +0.4 | +2.7 | |
| Chelsea @ EVE | +0.4 | −0.2 | |
| Liverpool @ BRE | +0.4 | 0.0 | |
| Manchester City v IPS | +0.2 | +0.5 | |
| Coventry @ TOT | 0.0 | −0.3 | |
| Leeds v MUN | −0.3 | +0.1 | |
| Aston Villa @ NEW | −0.4 | +0.2 | |
| Brentford v LIV | −0.5 | −0.7 | |
| Manchester United @ LEE | −0.7 | −0.7 | |
| Arsenal @ NFO | −0.8 | −0.6 | |
| Nottingham Forest v ARS | −0.9 | +0.1 | |
| Fulham v HUL | −0.9 | −0.5 | |
| Hull @ FUL | −0.9 | −0.7 | |
| Sunderland @ BOU | −1.1 | −1.0 | |
| Crystal Palace @ BHA | −1.1 | −0.3 | |
| Everton v CHE | −1.2 | −0.8 | |
| Newcastle v AVL | −1.7 | −0.7 |
Signal: chances
Bournemouth made 4.3 xG against 1.6 expected and scored 2. Chance creation running ahead of the market is the kind of result that can move a rating.
Noise: finishing
Brighton scored 5 from 2.3 xG (expected 1.9). A big week for their attackers, but finishing streaks rarely carry into the next match.
The season so far: where the market has been wrong
Season xG compared with the goals the closing prices expected, per team. These are the gaps the ratings can still close.
Creating more than expected
- Brighton+4.5
- Sunderland+3.3
- Bournemouth+3.0
- Nottingham Forest+2.8
Creating less than expected
- Tottenham−3.6
- Newcastle−3.2
- Chelsea−2.2
- Liverpool−2.1
Allowing less than expected
- Hull+3.4
- Bournemouth+2.7
- Coventry+2.5
- Nottingham Forest+1.9
Allowing more than expected
- Sunderland−3.5
- Crystal Palace−3.3
- Manchester City−2.9
- Newcastle−1.9
What changes in the model
Team ratings come from bookmakers’ closing prices, not results, so they move when the market’s view moves: results backed by chances tend to move it, finishing luck mostly doesn’t.
Biggest movers since the deadline, and what that does to their next five fixtures (GW8–12):
| Manchester United | 1.87 → 1.75 | 1.29 → 1.29 | 9.3 → 8.7 | 7.2 → 7.1 |
|---|---|---|---|---|
| Brighton | 1.49 → 1.57 | 1.37 → 1.34 | 7.4 → 7.9 | 7.7 → 7.6 |
| Sunderland | 1.14 → 1.11 | 1.42 → 1.35 | 6.0 → 5.8 | 6.9 → 6.6 |
| Arsenal | 1.91 → 1.98 | 0.88 → 0.86 | 9.9 → 10.4 | 4.6 → 4.5 |
| Leeds | 1.34 → 1.34 | 1.37 → 1.29 | 6.9 → 6.9 | 6.9 → 6.4 |
| Bournemouth | 1.45 → 1.40 | 1.44 → 1.40 | 7.5 → 7.2 | 7.2 → 6.9 |
Attack and conceding are goals per match against an average side, at the deadline → now.
2 of 5
Positions and playing time
22 players we had at 70%+ to start didn't. 2 were flagged by FPL beforehand. Cherki (84%, 38% owned) was the costly one. B.Fernandes played right wing instead of attacking midfield.
- start forecast error (Brier) over 667 players; 0.25 is a coin flip
- 0.089
- at 70%+ to start who didn’t (2 flagged by FPL)
- 22
- started from under 30%
- 22
- played a different line from the model’s
- 13
Expected to start, didn’t
| Cherki Manchester City | 84% | 0 |
|---|---|---|
| Gvardiol Manchester City | 73% | 29 |
| Schade Brentford | 71% | 45 |
| Semenyo Manchester City | 96% | 13 |
| Gibbs-White Nottingham Forest | 73% | 0 |
| Cunha Manchester United | 94% | 6 |
Started against the odds
| Maddison Tottenham | 0% | 9 |
|---|---|---|
| Dominguez Nottingham Forest | 4% | 3 |
| Bouaddi Manchester City | 1% | 2 |
| Norton-Cuffy Hull | 2% | 2 |
| Tosin Tottenham | 2% | 2 |
| Walton Ipswich | 3% | 2 |
Played a different line
| Model | Played | |
|---|---|---|
| B.Fernandes Manchester United | AM | RW |
| Palmer Chelsea | RW | AM |
| Mbeumo Manchester United | RW | CF |
| Wissa Newcastle | AM | CF |
| O'Reilly Manchester City | CM | LB |
| Belloumi Hull | RWB | RW |
Plus 29 who changed position within the same line.
Taken off early (4)
| Martinez Manchester United | 71′ | 45′ |
|---|---|---|
| Willock Newcastle | 78′ | 55′ |
| Timber Crystal Palace | 75′ | 45′ |
| Fernandez-Pardo Newcastle | 76′ | 55′ |
What changes in the model
GW8 start chances that moved most since the deadline, and why:
| Why | |||
|---|---|---|---|
| Martinez Chelsea | 96% | 0% | Head injury - Unknown return date |
| Semenyo Manchester City | 96% | 0% | Knee injury - Unknown return date |
| Janelt Brentford | 93% | 0% | Suspended until 31 Oct |
| Isak Liverpool | 23% | 92% | fit again (FPL flag cleared) |
| Gakpo Liverpool | 22% | 86% | fit again (FPL flag cleared) |
| Elanga Newcastle | 0% | 58% | Unspecified injury - Expected back 25 Oct |
| Mykolenko Everton | 38% | 81% | fit again (FPL flag cleared) |
| Ndiaye Manchester City | 50% | 90% | start record 6 of 7 |
WatchingB.Fernandes at right wing: one match so far. A new position is recorded as one the player also plays; if it repeats, his role (and goal share) changes.
WatchingPalmer at attacking midfield: one match so far. A new position is recorded as one the player also plays; if it repeats, his role (and goal share) changes.
WatchingMbeumo at centre-forward: one match so far. A new position is recorded as one the player also plays; if it repeats, his role (and goal share) changes.
3 of 5
Injuries, suspensions and congestion
Liverpool lost Szoboszlai and Frimpong. 9 teams played midweek before GW7; their likely starters still made 69 of the 71 starts we expected.
New or worse since the deadline
| Now | |
|---|---|
| Haaland Manchester City | doubtfulUnspecified injury - 75% chance of playing |
| Szoboszlai Liverpool | injuredHamstring injury - Unknown return date |
| Semenyo Manchester City | injuredKnee injury - Unknown return date |
| O'Reilly Manchester City | doubtfulBack injury - 75% chance of playing |
| Brobbey Sunderland | doubtfulHamstring injury - 75% chance of playing |
| Martinez Chelsea | injuredHead injury - Unknown return date |
| Manzambi Aston Villa | doubtfulCalf injury - 75% chance of playing |
| Janelt Brentford | suspendedSuspended until 31 Oct |
And 5 more.
Back
Schade, Isak, Gakpo, Mykolenko, Branthwaite, Robinson, Mateta
Sent off
Janelt (Brentford), Yarmoliuk (Brentford), Sessegnon (Fulham), Danso (Sunderland)
Fixture congestion
Before GW7, 9 teams had a cup or European match:
| Aston Villa | Wed 14 Oct: Fenerbahçe (h) Champions League | Sat 17:30: Newcastle (a) | 3 days |
|---|---|---|---|
| Bournemouth | Thu 15 Oct: Sturm Graz (h) Europa League | Sun 14:00: Sunderland (h) | 3 days |
| Brighton | Thu 15 Oct: Kauno Žalgiris (h) Conference League | Sun 14:00: Crystal Palace (h) | 3 days |
| Crystal Palace | Thu 15 Oct: Lyon (a) Europa League | Sun 14:00: Brighton (a) | 3 days |
| Liverpool | Wed 14 Oct: LASK (a) Champions League | Sat 15:00: Brentford (a) | 3 days |
| Manchester City | Wed 14 Oct: Paris Saint-Germain (h) Champions League | Sat 15:00: Ipswich (h) | 3 days |
| Sunderland | Thu 15 Oct: Torreense (a) Europa League | Sun 14:00: Bournemouth (a) | 3 days |
| Arsenal | Tue 13 Oct: Lille (h) Champions League | Sun 16:30: Nottingham Forest (a) | 5 days |
| Manchester United | Tue 13 Oct: Atletico Madrid (a) Champions League | Sun 14:00: Leeds (a) | 5 days |
Did they rotate?
| Teams | Likely starters | Expected starts | Started | Rate v expected |
|---|---|---|---|---|
| Played midweek (9) | 78 | 71.2 | 69 | −3% |
| Didn’t (11) | 99 | 88.0 | 86 | −2% |
One gameweek is a small sample; this builds week by week into the evidence for whether to model congestion.
Before GW8, 9 teams play midweek: rest days and the teams affected are in the GW8 preview.
What changes in the model
- Injuries and suspensions feed straight into start chances (see the start-chance moves above).
- Midweek matches are not modelled. Start chances ignore rest days. Each reflection now compares how often likely starters at teams with a midweek match actually started against teams without one; if they start noticeably less over several weeks, that is the case for adding it to the model.
Price changes since the deadline (16)
| João Pedro Chelsea | £7.8m | £7.9m | 72.8% |
|---|---|---|---|
| Calafiori Arsenal | £5.9m | £6.0m | 53.8% |
| Rogers Chelsea | £7.9m | £8.0m | 48% |
| Groß Brighton | £6.1m | £6.2m | 45.3% |
| De Cuyper Brighton | £5.2m | £5.3m | 41% |
| Cherki Manchester City | £7.9m | £8.0m | 37.7% |
| Virgil Liverpool | £6.5m | £6.4m | 15.3% |
| Wissa Newcastle | £6.2m | £6.1m | 14.1% |
| Gibbs-White Nottingham Forest | £8.0m | £7.9m | 11.4% |
| Diop Ipswich | £4.0m | £3.9m | 11% |
| Tavernier Bournemouth | £6.1m | £6.2m | 9.2% |
| Manzambi Aston Villa | £5.9m | £6.0m | 2.8% |
| Doku Manchester City | £7.4m | £7.5m | 2.4% |
| Aina Nottingham Forest | £4.4m | £4.3m | 1.8% |
| Robinson Fulham | £4.5m | £4.4m | 1.3% |
| Khusanov Manchester City | £5.4m | £5.3m | 1.2% |
4 of 5
Set pieces
No change to FPL's set-piece takers. No penalties missed. Our penalty taker differs from FPL's at 1 club.
No change in FPL’s first-choice set-piece takers since the deadline.
- penalties missed
- 0
- clubs where our penalty taker differs from FPL’s
- 1
- our set-piece hierarchy last reviewed
- Sat 19 Sept
What changes in the model
The projection gives set-piece chances to the takers in our own hierarchy (left and right corners, free kicks, penalties), which is reviewed by hand. FPL’s order is a cross-check, not an input.
Penalty takers that differ: Coventry (FPL Rudoni; ours Wright).
We can’t see who took each penalty. FPL’s data doesn’t say which goals were penalties, so a new taker only shows up when FPL changes its order. A source for penalty events would let us spot it sooner.
5 of 5
Captaincy
Haaland scored 13; the underlying-form pick, Saka, 3; last week's top scorer, João Pedro, 2. Since GW2, captaining the premium every week has 112 captain points; underlying form 28 and chasing last week's top scorer 32.
This week: the range at the deadline, and where each landed
| Candidate | Range | ≤2 | 10+ | Pts |
|---|---|---|---|---|
| Haaland most captained | 3% | 35% | 13 | |
| Saka underlying form | 9% | 16% | 3 | |
| João Pedro last week's top scorer | 20% | 11% | 2 |
≤2: chance of 2 points or fewer. 10+: chance of 10 or more (before doubling). From 20,000 simulated gameweeks using the projections at the deadline.
Season so far: return and volatility (GW2–7)
| Premium every week | 112 | 9.3 | 3.1 | 6 | 13 | 0 | 2 |
|---|---|---|---|---|---|---|---|
| Underlying form | 28 | 2.3 | 1.0 | 1 | 4 | 4 | 0 |
| Last week's top scorer | 32 | 2.7 | 2.3 | 0 | 7 | 4 | 0 |
| Hindsight best | 180 | 15.0 | 4.3 | 12 | 23 | 0 | 6 |
- Premium every week: FPL's most-captained player each gameweek.
- Underlying form: most expected goal involvements (xGI) over the previous 4 gameweeks, among midfielders and forwards who cost £7.0m+ at the start of the season, played 180+ minutes in them and started their last match; never the premium.
- Last week's top scorer: most points in the previous gameweek among the same players; never the premium.
- Hindsight best: the top scorer among those players and the premium (a ceiling, not a strategy).
- Model's top projection: the highest projection at each deadline (saved from GW6).
- Per week, worst and best are points before doubling; spread is the standard deviation; a blank is 2 points or fewer.
The model’s top projection has been saved since GW6: 30 captain points so far against 38 for the premium in the same weeks.
Over four seasons
The most-captained player blanked in 27.7% of gameweeks from 2022/23 to 2025/26 and was the best choice in only 19.9%, yet scored the most over the four seasons. A few weeks of this season decide very little. See should you always captain the top pick and should you always captain Haaland.
Week by week
| Premium | xGI form | Last top scorer | Model | Best | |
|---|---|---|---|---|---|
| 1 | Haaland 2 | – | – | – | Palmer 13 |
| 2 | Haaland 13 | Szoboszlai 4 | Palmer 7 | – | B.Fernandes 23 |
| 3 | Haaland 9 | B.Fernandes 2 | B.Fernandes 2 | – | Isak 13 |
| 4 | Haaland 9 | B.Fernandes 2 | Isak 2 | – | João Pedro 12 |
| 5 | Haaland 6 | B.Fernandes 2 | João Pedro 0 | – | Semenyo 17 |
| 6 | Haaland 6 | B.Fernandes 1 | Semenyo 3 | Saka 2 | Doku 12 |
| 7 | Haaland 13 | Saka 3 | João Pedro 2 | Haaland 13 | Haaland 13 |
How this was worked out
Generated from the data on Tue 20 Oct; nothing is written by hand. Expected goals at the deadline: deadline snapshot (market goals). Start chances and roles are the ones the projections had at the deadline (Sat 17 Oct). xG is FPL’s. Team ratings come from bookmakers’ closing prices, including this gameweek’s. Players listed are owned by 1%+ or among the 50 highest projected for the next five gameweeks.
FPL weekly email
One email before each deadline: our top captains, flagged players, price changes and the gameweek preview.