Back Bruno, buy the bad luck
GW8 is the first deadline brief: everything you need before Friday's deadline on one page, built from the same numbers as the projections and planner. Three calls up top, then the evidence. Every brief from now on starts by grading the previous week's calls, so you can see whether to trust them.
The brief three calls for GW8
The model's highest projection of the week, 0.4 clear of Saka, at home to Bournemouth in the fixture with the most expected goals. Its boldest call too: five straight returns of 1 or 2 points.
Bournemouth are the unluckiest side in the league by expected points, and he is their most dangerous midfielder (2.89 xG, four goals). Fifth-best midfielder over five weeks at £6.2m.
90 minutes in every game, £4.0m, and the most projected points per pound of any player in the game. The bench slot that pays for everything else.
Captaincy 20,000 simulated GW8s
| Player | expected 1-in-10 ceiling | xPts | Pens | Ceiling | 10+ pts | Owned |
|---|---|---|---|---|---|---|
| MID B.FernandesMUN · BOU (H) | 6.4 | 0.6 | 11 | 16% | 29% | |
| MID SakaARS · EVE (H) | 6.0 | 0.5 | 12 | 15% | 17% | |
| DEF GabrielARS · EVE (H) | 5.8 | – | 9 | 7% | 21% | |
| MID MbeumoMUN · BOU (H) | 5.4 | – | 10 | 12% | 20% | |
| MID PalmerCHE · TOT (H) | 5.3 | 0.6 | 10 | 12% | 26% | |
| DEF CalafioriARS · EVE (H) | 4.9 | – | 9 | 8% | 54% | |
| MID RiceARS · EVE (H) | 4.6 | – | 8 | 4% | 11% | |
| FWD Haaland 75% fitMCI · AVL (A) | 4.4 | 0.2 | 10 | 12% | 77% |
Fernandes leads by 0.4, and it is worth being clear about why, because the form says the opposite. He scored 23 points in GW2 and has managed 1 or 2 in each of the five games since, and 177,000 managers have sold him this week. The model now splits penalties out of xG, so his 6.4 breaks down cleanly: 1.2 from open-play goals, 0.6 from penalties, 1.4 from assists, and the rest from minutes and bonus. That's also why he sits ahead of Mbeumo (5.4), who has the bigger open-play goal threat (2.1) but is second in line for penalties and creates less. Add a home game against Bournemouth in the match with the most expected goals of the week (United's 1.89 is the highest of any side) and he comes out on top. The model picked him once already, in GW6, and he scored 1. Treat this as its highest-conviction call, not a safe one.
Haaland is the captain most people will choose, and he carries a 75% flag. His 4.4 already includes the one-in-four chance he doesn't play. Fit, he projects at roughly 5.9 at Villa Park, still below Fernandes but close enough that captaining him to protect your rank is reasonable if he's named in the XI. If he isn't fit, Saka (6.0 at home to Everton, and Arsenal's penalty taker) is the clean alternative. Gabriel at 5.8 is the defender's version of the same Arsenal fixture: the highest clean-sheet odds of the week, and a 1-in-10 ceiling of about 9 points.
The fixtures model xG · clean-sheet odds
Arsenal at home to Everton is the standout: 44% clean-sheet odds, the only side above 35%, and 1.84 expected goals. United (1.89 xG) and Chelsea (1.83 at home to Spurs) have the best attacking fixtures. The model rates Liverpool v Brighton as close to a coin toss, and it rates City as the stronger side at Villa Park by some distance. That 34% clean-sheet chance for City is the second best of the week. Forest are the model's pick of the away sides. They have 1.78 expected goals at Ipswich on Friday night, more than City at Villa Park.
The model's XI 3-5-2 · £100.0m
Best 15 under £100.0m for the next five gameweeks, picked by the optimiser on the projections page. Expected GW8 total with the armband: 59.6.
The optimiser's squad leans hard on two fixtures: three Arsenal players for the Everton game and two United midfielders for Bournemouth. It has no Haaland, and that's the price of £15.7m for a doubtful player, not a judgement on his ability. The rest of the XI has two Brighton players and one each from City, Bournemouth, Brentford and Leeds, mostly mid-priced players with secure minutes. If you're on a wildcard, start here and swap in the template players you want to keep.
Transfer targets GW8–12 projection
| Player | GW8–12 | Per £m | GW8 | Next five (model difficulty) |
|---|---|---|---|---|
| GKP | ||||
| RayaARS · £6.1m · 44% | 20.1 | 3.30 | 4.2 | EVEHLIVAHULHNEWAMCIH |
| Trafford differentialLEE · £5.0m · 5.9% | 16.0 | 3.20 | 3.2 | SUNABOUATOTHCHEACOVH |
| Suzuki differentialAVL · £5.0m · 2.0% | 16.0 | 3.20 | 3.0 | MCIHFULHMUNASUNHIPSA |
| Verbruggen valueBHA · £4.6m · 23% | 15.4 | 3.34 | 3.0 | LIVAMCIABREHHULANEWH |
| Leno valueFUL · £4.5m · 4.2% | 15.0 | 3.33 | 3.2 | COVAAVLANEWHMCIABOUH |
| DEF | ||||
| GabrielARS · £7.9m · 21% | 27.7 | 3.51 | 5.8 | EVEHLIVAHULHNEWAMCIH |
| CalafioriARS · £6.0m · 54% | 23.0 | 3.84 | 4.9 | EVEHLIVAHULHNEWAMCIH |
| GuéhiMCI · £6.0m · 18% | 20.6 | 3.43 | 4.5 | AVLABHAHNFOAFULHARSA |
| Thomas valueCOV · £4.0m · 9.3% | 18.4 | 4.61 | 3.8 | FULHSUNHEVEACRYHLEEA |
| Justin valueLEE · £4.5m · 2.1% | 18.5 | 4.11 | 3.7 | SUNABOUATOTHCHEACOVH |
| MID | ||||
| B.FernandesMUN · £11.8m · 29% | 30.9 | 2.62 | 6.4 | BOUHCHEAAVLHLIVABREH |
| SakaARS · £9.6m · 17% | 28.6 | 2.98 | 6.0 | EVEHLIVAHULHNEWAMCIH |
| MbeumoMUN · £7.8m · 20% | 26.2 | 3.35 | 5.4 | BOUHCHEAAVLHLIVABREH |
| Rudoni valueCOV · £4.9m · 0.3% | 19.6 | 3.99 | 3.9 | FULHSUNHEVEACRYHLEEA |
| Tavernier valueBOU · £6.2m · 9.1% | 23.3 | 3.76 | 4.3 | MUNALEEHIPSANFOHFULA |
| FWD | ||||
| HaalandMCI · £15.7m · 77% | 26.2 | 1.67 | 4.4 | AVLABHAHNFOAFULHARSA |
| Calvert-LewinLEE · £6.0m · 23% | 21.8 | 3.63 | 4.3 | SUNABOUATOTHCHEACOVH |
| Thiago differentialBRE · £7.8m · 8.8% | 21.5 | 2.76 | 4.3 | HULANFOHBHAAEVEHMUNA |
| Barry valueEVE · £5.7m · 8.4% | 18.7 | 3.29 | 3.1 | ARSANEWACOVHBREALIVH |
| Gonzalo valueFUL · £6.1m · 6.9% | 19.8 | 3.25 | 4.0 | COVAAVLANEWHMCIABOUH |
The value picks (the value tag) are the best projection per pound among players already in the top 20 at their position, so they aren't just cheap. Thomas of Coventry is the clearest case. He has played every minute bar 13 this season and projects 18.4 points over five weeks at £4.0m, more per pound than anyone in the game. Rudoni at £4.9m is the midfield version, on penalties for Coventry, though he's started six of seven. Arsenal's next five are the easiest in the league by the model's ratings, alongside City's. That's why Gabriel and Calafiori sit at the top of the defenders despite their prices.
Crowd vs model this week's transfers
| Most bought | In | Model rank | Verdict |
|---|---|---|---|
| GroßBHA · £6.2m | 161k | #6 MID | Agree |
| CherkiMCI · £8.0m | 159k | #36 MID | Model is cooler |
| TavernierBOU · £6.2m | 145k | #5 MID | Agree |
| De CuyperBHA · £5.3m | 139k | #4 DEF | Agree |
| RogersCHE · £8.0m | 119k | #10 MID | Agree |
| Most sold | Out | Model rank | Verdict |
|---|---|---|---|
| SzoboszlaiLIV · £6.8m | 252k | #111 MID | Agree: injured |
| SemenyoMCI · £8.3m | 213k | #120 MID | Agree: injured |
| B.FernandesMUN · £11.8m | 177k | #1 MID | Model disagrees |
| WissaNEW · £6.1m | 157k | #9 FWD | Model disagrees |
| GvardiolMCI · £5.8m | 89k | #68 DEF | Agree |
Model rank = position among players at the same position on the GW8–12 projection. Transfers are this gameweek's totals so far, from the FPL API.
Two disagreements stand out. Cherki is the second most-bought player of the week, but the model ranks him 36th among midfielders, because it gives him only a 65% chance of starting. He has started five of City's seven games, and Guardiola rotates. On the other side, Wissa is being sold after one goal from 2.97 xG. That's the profile of a forward due a goal, not one to drop, and the model has him ninth among forwards for the next five. Fernandes is the third, covered above. The crowd agrees with the model on Groß, Tavernier, De Cuyper and Rogers.
Fitness flags 5%+ owned
- Haaland MCI · 77% owned75%Unspecified injury - 75% chance of playing
- Szoboszlai LIV · 20% ownedOutHamstring injury - Unknown return date
- Semenyo MCI · 13% ownedOutKnee injury - Unknown return date
- Gibbs-White NFO · 12% owned75%Muscular injury - 75% chance of playing
- O'Reilly MCI · 9.2% owned75%Back injury - 75% chance of playing
- Shaw MUN · 5.7% owned75%Unspecified injury - 75% chance of playing
- Brobbey SUN · 5.6% owned75%Hamstring injury - 75% chance of playing
- Mendy HUL · 5.5% ownedOutGroin injury - Expected back 31 Oct
Haaland's 75% is the flag that matters: 77% of managers own him. City's pre-match press conference is the one to watch. Szoboszlai and Semenyo have no return date, so the transfers out are right. Both are 0 in the model until FPL changes their status.
Luck check points minus Understat xPts, 2026/27
Understat's expected points turn each game's chances into the points a team "should" have taken. Forest (5 points from 12.6 expected) and Bournemouth (6 from 12.3) are the two sides the table is most unfair to. By chance quality, they're the fourth and fifth best teams in the league, not 18th and 15th. Forest's next five are hard (the joint-worst run on the model's ratings). Bournemouth's are the third easiest, which is why Tavernier is the buy rather than a Forest attacker. At the other end, Chelsea are fourth in the table but 12th by expected points, and City's 21 points from 14.5 expected suggests their title-race lead is partly finishing. Neither is a reason to sell Palmer or Haaland, but it is a reason not to pay extra for Chelsea's defenders.
Track record live since GW6
| GW | Model error | FPL error | Rank corr. | Model captain | FPL's captain |
|---|---|---|---|---|---|
| GW6 | 1.94 | 2.21 | 0.32 | B.Fernandes 7.2 → 1 | Groß → 6 |
| GW7 | 1.97 | 2.35 | 0.37 | Haaland 7.7 → 13 | De Cuyper → 6 |
Two weeks of live tracking is a small sample, so read this as an early check rather than a verdict. The model's average error per player has been lower than FPL's own expected-points figure in both weeks, about a third of a point better. Its ranking of players has been better on average too, though FPL's was slightly better in GW6. Its captains have scored 1 (Fernandes, GW6) and 13 (Haaland, GW7). The full week-by-week table is on the track record page.
Method: projections are the site's own model (non-penalty xG plus a separate penalty term, xA, clean-sheet and minutes rates, shrunk toward last season and price; details on the metrics page), generated Wed 21 Oct 21:40. Captaincy uses 20,000 simulations of GW8; the ceiling is the 90th percentile. Fixture difficulty is the model's 1–5 blend of xG ratings and FPL's own. Luck check uses Understat expected points. Check team news before the deadline: the model only knows what FPL's availability flags said at the time.