Free FPL Tools: Clean Sheet Probability, Fixture Difficulty and More
The model's squad is frozen before every deadline. We play it in FPL as entry 116920, with our own chip calls and the occasional lineup change, and that's the team to beat in the Beat the Model mini-league (code jgi6j9). Season winner gets a year of GoalIQ Premium, free: one prize, decided by the mini-league table when the season ends.
Team news right now
76 players are ruled out and 40 are doubtful for the next deadline, taken from the official Fantasy Premier League status feed. Most owned among them: João Pedro (CHE), Palmer (CHE), Isak (LIV).
The full list is free and sorted by ownership, and it carries the model's projected points where we have them. The number you see is what the model expects including the doubt, not what the player would score if fully fit: FPL team news.
Gameweek 6 clean sheet probabilities
Model clean sheet probability for the 20 Premier League teams with a fixture in Gameweek 6 (Saturday 10 October 2026). FDR is GoalIQ's model fixture difficulty for that match, 1 easiest to 5 hardest. Next 6 CS% averages the same probability over the run in the grid below, so one good fixture and a good run are not the same column.
| Team | Clean sheet % | Next opponent | FDR | Next 6 CS% | Squad turnover |
|---|---|---|---|---|---|
| Arsenal6% turnover | 46.5% | Leeds (H)projected goals ARS 1.92 v LEE 0.77 | 1 | 44% | 6% |
| Newcastle United25% turnover | 39.9% | Coventry (A)projected goals NEW 1.20 v COV 0.92 | 2 | 25% | 25% |
| Hullrating from 5 matches | 38.6% | Everton (H)projected goals HUL 1.14 v EVE 0.95 | 2 | 30% | 5 matches |
| Manchester United7% turnover | 38.3% | Tottenham (H)projected goals MUN 2.37 v TOT 0.96 | 1 | 26% | 7% |
| Everton20% turnover | 31.9% | Hull (A)projected goals EVE 0.95 v HUL 1.14 | 3 | 24% | 20% |
| Brighton26% turnover | 30.8% | Sunderland (A)projected goals BHA 1.57 v SUN 1.18 | 2 | 26% | 26% |
| Coventryrating from 5 matches | 30.3% | Newcastle United (H)projected goals COV 0.92 v NEW 1.20 | 3 | 27% | 5 matches |
| Ipswichrating from 5 matches | 27.3% | Fulham (H)projected goals IPS 1.38 v FUL 1.30 | 3 | 16% | 5 matches |
| Manchester City18% turnover | 26.7% | Liverpool (A)projected goals MCI 1.66 v LIV 1.32 | 2 | 36% | 18% |
| Aston Villa36% turnover | 25.5% | Brentford (H)projected goals AVL 1.53 v BRE 1.37 | 3 | 21% | 36% |
| Fulham22% turnover | 25.2% | Ipswich (A)projected goals FUL 1.30 v IPS 1.38 | 3 | 26% | 22% |
| Nottingham Forest12% turnover | 23.7% | Crystal Palace (A)projected goals NFO 1.62 v CRY 1.44 | 3 | 20% | 12% |
| Brentford7% turnover | 21.7% | Aston Villa (A)projected goals BRE 1.37 v AVL 1.53 | 4 | 24% | 7% |
| Sunderland9% turnover | 20.9% | Brighton (H)projected goals SUN 1.18 v BHA 1.57 | 4 | 25% | 9% |
| Crystal Palace20% turnover | 19.8% | Nottingham Forest (H)projected goals CRY 1.44 v NFO 1.62 | 4 | 25% | 20% |
| Liverpool17% turnover | 19% | Manchester City (H)projected goals LIV 1.32 v MCI 1.66 | 4 | 21% | 17% |
| Chelsea28% turnover | 18.5% | Bournemouth (H)projected goals CHE 1.88 v BOU 1.69 | 3 | 23% | 28% |
| Bournemouth16% turnover | 15.3% | Chelsea (A)projected goals BOU 1.69 v CHE 1.88 | 4 | 24% | 16% |
| Leeds13% turnover | 14.7% | Arsenal (A)projected goals LEE 0.77 v ARS 1.92 | 5 | 23% | 13% |
| Tottenham18% turnover | 9.4% | Manchester United (A)projected goals TOT 0.96 v MUN 2.37 | 5 | 22% | 18% |
Team strengths are fitted on 2025/26 and 2026/27 results, 5 gameweeks of 2026/27 included. Coventry, Hull, Ipswich came up this summer, so their ratings rest on 5 Premier League matches each and move a lot with every result. The numbers sharpen as 2026/27 results arrive.
This is Gameweek 6 alone. GoalIQ Premium projects clean sheets and expected points for every gameweek in the window, so you can see whether a defender is worth buying for one weekend or for the run.
Fixture difficulty for the next six gameweeks
Each cell is the opponent, venue and the model's clean sheet probability for that match. Model FDR (1 easiest, 5 hardest) is in the cell tooltip on desktop. Model-derived, not the official FPL difficulty.
Long-range fixture difficulty
Fixture difficulty only, based on today's ratings. Clean sheet % appears as each gameweek moves closer. Lower is easier. Each column averages the model's difficulty over six gameweeks, so this shows where the swings are, not a number for any single match.
| Team | GW12-17 | GW18-23 | GW24-29 | GW30-35 | GW36-38 |
|---|---|---|---|---|---|
| Arsenal | 1.3 | 1.3 | 1.5 | 1.0 | 1.0 |
| Manchester City | 2.0 | 1.7 | 1.0 | 1.3 | 1.0 |
| Manchester United | 2.5 | 2.5 | 1.8 | 2.0 | 1.0 |
| Brighton | 1.7 | 2.8 | 2.0 | 2.0 | 3.0 |
| Liverpool | 2.2 | 2.3 | 2.2 | 2.8 | 3.0 |
| Bournemouth | 2.7 | 2.2 | 1.8 | 3.3 | 3.0 |
| Brentford | 3.3 | 2.7 | 2.3 | 2.2 | 3.3 |
| Hull | 3.2 | 2.3 | 3.3 | 2.8 | 2.3 |
| Chelsea | 2.5 | 2.5 | 3.5 | 3.2 | 2.7 |
| Leeds | 2.7 | 2.5 | 2.7 | 3.5 | 3.3 |
| Everton | 2.8 | 2.8 | 3.2 | 2.7 | 3.7 |
| Aston Villa | 2.5 | 3.7 | 3.5 | 3.3 | 2.7 |
| Nottingham Forest | 3.5 | 3.7 | 2.8 | 3.7 | 2.7 |
| Newcastle United | 3.7 | 3.7 | 3.7 | 2.3 | 3.3 |
| Sunderland | 3.8 | 3.5 | 3.0 | 3.3 | 4.3 |
| Fulham | 3.8 | 3.7 | 4.2 | 3.3 | 3.3 |
| Crystal Palace | 4.0 | 3.8 | 3.2 | 3.7 | 4.0 |
| Tottenham | 3.7 | 3.8 | 4.3 | 4.0 | 3.3 |
| Ipswich | 3.8 | 4.3 | 4.7 | 4.5 | 4.3 |
| Coventry | 4.7 | 4.2 | 4.8 | 4.8 | 4.3 |
Clean sheet scale: 10% 22% 34% 46% 58% (low CS% = hard fixture, high CS% = easy). H home, A away.
The grid says which fixtures turn. Rate my team is free and needs no account: it names the line that is costing you and gives you a captain. Premium is what comes after that, the transfer chains over the run with the hit priced in.
The model publishes its prediction record
The GoalIQ club model's predictions for domestic league and Champions League matches are logged before kick-off and never edited after it. Across the 553 completed club matches, it called the result correctly in 48.1% of them. In the 404 club matches that did not end in a draw, it called the result right 65.8% of the time (266 of 404). The model always names a side, so every draw counts as a miss and that 65.8% is the same 266 hits over a smaller number of matches. World Cup 2026 predictions came from a separate national-team model and have their own row in the prediction record.
Source: GoalIQ prediction log, updated 7 October 2026. The full log, match by match with every miss included, is on the prediction record page.
Gameweek calls, logged and scored
The captain of the model's own FPL squad and the four picks on the weekly standouts card (captain pick, ceiling, safest pick, boom or bust) go into a log with a timestamp. The row follows the latest projection until the FPL deadline and never changes after it. The Logged column shows the last write before the deadline. Once the gameweek has been played, each call is scored with official FPL points. A hit means the player did what the call said: 10 or more points for the captain pick and the boom-or-bust pick, the ceiling number or more for ceiling, 3 or more for the safest pick. Provisional rows wait for FPL to confirm bonus points.
| GW | Call | Player | Logged | What it said | Points | Result |
|---|---|---|---|---|---|---|
| GW5 | Model squad captain | Haaland (MCI) | 18 Sep 16:49 UTC, 40 min before the deadline | captain, points doubled | 6 | 12 as captain |
| GW5 | Captain pick | Haaland (MCI) | 18 Sep 16:49 UTC, 40 min before the deadline | 22% chance of 10+ | 6 | miss |
| GW5 | Ceiling | Thiago (BRE) | 18 Sep 16:49 UTC, 40 min before the deadline | ceiling 11 pts | 5 | miss |
| GW5 | Safest pick | Stach (LEE) | 18 Sep 16:49 UTC, 40 min before the deadline | 82% chance of 3+ | 5 | hit |
| GW5 | Boom or bust | Palmer (CHE) | 18 Sep 16:49 UTC, 40 min before the deadline | 9% chance of 10+ | 2 | miss |
| GW4 | The squad frozen for GW4 in data/model_squad_frozen is not the squad that played. It changes 8 of the 15 players from GW3 while listing no transfers, because the freeze rebuilt it from scratch by mistake. The squad that played GW4 kept its GW3 players and captained B.Fernandes, the captain in the row below; its picks are public at FPL entry 116920. | |||||
| GW4 | Model squad captain | B.Fernandes (MUN) | 12 Sep 11:50 UTC, 39 min before the deadline | captain, points doubled | 2 | 4 as captain |
| GW4 | Captain pick | B.Fernandes (MUN) | 12 Sep 11:50 UTC, 39 min before the deadline | 11% chance of 10+ | 2 | miss |
| GW4 | Ceiling | Haaland (MCI) | 12 Sep 11:50 UTC, 39 min before the deadline | ceiling 10 pts | 9 | miss |
| GW4 | Safest pick | Stach (LEE) | 12 Sep 11:50 UTC, 39 min before the deadline | 83% chance of 3+ | 3 | hit |
| GW4 | Boom or bust | Khalaili (CRY) | 12 Sep 11:50 UTC, 39 min before the deadline | 10% chance of 10+ | 7 | miss |
| GW3 | The squad played a Triple Captain in GW3. Its picks are public at FPL entry 116920. | |||||
| GW3 | Model squad captain | Haaland (MCI) | 4 Sep 16:43 UTC, 46 min before the deadline | captain, points tripled | 9 | 27 as captain |
| GW3 | Captain pick | Haaland (MCI) | 4 Sep 16:43 UTC, 46 min before the deadline | 31% chance of 10+ | 9 | miss |
| GW3 | Ceiling | Isak (LIV) | 4 Sep 16:43 UTC, 46 min before the deadline | ceiling 11 pts | 13 | hit |
| GW3 | Safest pick | B.Fernandes (MUN) | 4 Sep 16:43 UTC, 46 min before the deadline | 82% chance of 3+ | 2 | miss |
| GW3 | Boom or bust | Calafiori (ARS) | 4 Sep 16:43 UTC, 46 min before the deadline | 7% chance of 10+ | 2 | miss |
| GW3 | Projected XI | 5-2-3 XI, captain Haaland (MCI) | 4 Sep 16:43 UTC, 46 min before the deadline | XI total, captain doubled, bench subs count | 56 | XI total |
| GW2 | The Model squad captain row below carries the captain from the squad frozen on 27 August, and that freeze was not redone after the model was rebuilt on deadline day. The squad that actually played GW2 wildcarded minutes after the row was logged and captained B.Fernandes; its picks are public at FPL entry 116920. The row was not rewritten before the deadline, so it is scored as logged. The four picks on the standouts card were recomputed with the rebuilt model. | |||||
| GW2 | Model squad captain | Guéhi (MCI) | 28 Aug 17:13 UTC, 17 min before the deadline | captain call, points doubled if captained | 2 | 4 if captained |
| GW2 | Captain pick | B.Fernandes (MUN) | 28 Aug 17:13 UTC, 17 min before the deadline | 15% chance of 10+ | 23 | hit |
| GW2 | Ceiling | Haaland (MCI) | 28 Aug 17:13 UTC, 17 min before the deadline | ceiling 10 pts | 13 | hit |
| GW2 | Safest pick | Stach (LEE) | 28 Aug 17:13 UTC, 17 min before the deadline | 81% chance of 3+ | 4 | hit |
| GW2 | Boom or bust | Guéhi (MCI) | 28 Aug 17:13 UTC, 17 min before the deadline | 8% chance of 10+ | 2 | miss |
Source: data/gw_calls.json in the public repository, with the squad the captain came from in data/model_squad_frozen. The commit history has the earlier versions of each row. The percentages are the same simulations as the 10+, Blank and Ceiling columns on the free expected points page, where a 3+ chance is 100 minus the Blank column on the free page (2 points or fewer, including not playing).
How far off the per player projections were
The projection is locked into a file before the deadline, so it can't be tidied up afterwards. Every player in the projection is in there, and all of them are graded against the points FPL gave them, the ones who never got on the pitch included. The figure is the mean absolute error, MAE: the average gap in points between the projection and what the player scored. Each gameweek number below links to that gameweek's own page, where the same comparison is open player by player. That page also states a second, higher figure for the players who actually took the pitch: the number in this table is lower because it includes the players who did not play, whose score was 0 and whose projection was already low. The freeze also stores FPL's own expected points for the coming gameweek (the ep_next field) and the player's FPL form, read from the FPL API at the moment of the freeze. A player counts in those columns only when all three numbers were frozen.
| GW | Players | GoalIQ xP | FPL ep_next | FPL form |
|---|---|---|---|---|
| GW5 | 482 | 1.62 | 1.59 | 1.61 |
| GW4 | 493 | 1.68 | 1.65 | 1.65 |
| GW3 | 501 | 1.54 | 1.69 | 1.63 |
| GW2 | 511 | 1.59 | not frozen | not frozen |
| GW1 | 490 | 1.76 | not frozen | not frozen |
The rows below pool every gameweek where all three numbers were frozen, so the columns can be read against each other.
| Group | Players | GoalIQ xP | FPL ep_next | FPL form |
|---|---|---|---|---|
| Did not play (0 minutes) | 566 | 0.97 | 0.55 | 0.49 |
| Played, blank (2 points or fewer) | 553 | 1.30 | 1.57 | 1.59 |
| Ticker (3 to 9 points) | 315 | 2.37 | 2.86 | 2.88 |
| Haul (10 or more points) | 42 | 8.64 | 8.12 | 8.12 |
| DEF | 496 | 1.89 | 1.91 | 1.93 |
| FWD | 155 | 1.67 | 1.66 | 1.65 |
| GKP | 164 | 1.46 | 1.09 | 1.06 |
| MID | 661 | 1.43 | 1.57 | 1.54 |
Source: data/fpl_xp_gw_accuracy.json in the public repository, with the frozen projections in data/fpl_xp_frozen, each written in a single commit dated before the deadline it was frozen for. Players outside the projection are not in these counts: FPL flagged them out (status i, s, u or n), or the model projects under 1.0 points for them across the next six gameweeks. A cell reads not frozen when that number was not saved for that gameweek; the FPL numbers start with the gameweek 3 freeze. On a row where the FPL numbers are frozen, all three come from the comparison block in the file, which counts only the players who had all three numbers, so that row shows fewer players than were frozen and a different GoalIQ figure than the gameweek's own mae.
What the top ranks own: effective ownership by rank tier
What the top of the table holds, not the whole game: managers sampled from three rank ranges after Gameweek 3, squads read from their Gameweek 4 picks. Chip points count towards rank, so the top 1k column is who led after Gameweek 3, not who is best. EO = mean multiplier x 100 (bench 0, playing 1, captain 2, triple captain 3), taken straight from the official picks payload. Sorted by the top 1k column.
| Player | EO top 1k, n=200 | EO top 10k, n=200 | EO top 100k, n=200 | Owned overall |
|---|---|---|---|---|
| Palmer (CHE) | 150.0% | 142.0% | 133.5% | 28.8% |
| João Pedro (CHE) | 116.0% | 113.5% | 115.5% | 74.1% |
| Calafiori (ARS) | 90.0% | 88.0% | 91.0% | 49.5% |
| Rogers (CHE) | 66.0% | 62.0% | 65.0% | 35.8% |
| Isak (LIV) | 64.0% | 58.0% | 49.5% | 24.5% |
| B.Fernandes (MUN) | 63.5% | 41.5% | 33.5% | 41.6% |
| De Cuyper (BHA) | 54.0% | 47.0% | 38.0% | 23.3% |
| Szoboszlai (LIV) | 47.5% | 50.5% | 50.0% | 38.6% |
| Haaland (MCI) | 45.0% | 72.5% | 88.5% | 71.1% |
| Raya (ARS) | 44.0% | 33.0% | 26.0% | 39.2% |
| Konsa (ARS) | 41.5% | 50.5% | 53.5% | 17.5% |
| Calvert-Lewin (LEE) | 36.5% | 28.5% | 32.0% | 21.4% |
| Mbeumo (MUN) | 29.0% | 26.0% | 26.5% | 22.9% |
| White (ARS) | 28.0% | 22.5% | 15.5% | 7.6% |
| Verbruggen (BHA) | 27.0% | 38.0% | 41.5% | 22.3% |
| Captain | top 1k, n=200 | top 10k, n=200 | top 100k, n=200 |
|---|---|---|---|
| Palmer (CHE) | 62.0% | 60.5% | 58.5% |
| João Pedro (CHE) | 18.5% | 17.0% | 18.0% |
| Haaland (MCI) | 7.5% | 9.0% | 10.5% |
| Isak (LIV) | 4.5% | 8.0% | 9.0% |
| Rogers (CHE) | 3.0% | 2.0% | 2.5% |
Each tier is a sample (top 1k n=200, top 10k n=200, top 100k n=200), so one manager in the top 1k sample is 0.5 points. Ranks after Gameweek 3, squads from Gameweek 4, so the two are not circular. Source: data/fpl_elite_ownership.json, generated 2026-09-12.
What the top ranks moved for Gameweek 4
Same managers, same sample, this time their transfers. The share is the part of the sample that made that move; a manager who kept the squad unchanged is counted too. Five per direction; where moves tie, the list cuts at five.
top 1k, n=200
Moved in: Rogers (CHE) 60.0%, Konsa (ARS) 37.5%, Hall (NEW) 34.0%, Palmer (CHE) 33.0%, De Cuyper (BHA) 28.0%.
Moved out: Mbeumo (MUN) 45.5%, B.Fernandes (MUN) 32.0%, Calvert-Lewin (LEE) 24.0%, Verbruggen (BHA) 23.0%, Ndiaye (MCI) 21.0%.
8.5% made no transfer and 3.0% took a points hit. Chips used, out of 200: wildcard 57, free hit 8, triple captain 24.
top 10k, n=200
Moved in: Rogers (CHE) 66.0%, Konsa (ARS) 52.5%, Palmer (CHE) 51.0%, Hall (NEW) 44.5%, De Cuyper (BHA) 30.5%.
Moved out: B.Fernandes (MUN) 52.0%, Mbeumo (MUN) 42.5%, Calvert-Lewin (LEE) 34.0%, Verbruggen (BHA) 27.0%, Maguire (MUN) 26.0%.
10.5% made no transfer and 4.0% took a points hit. Chips used, out of 200: wildcard 77, free hit 11, triple captain 11.
top 100k, n=200
Moved in: Rogers (CHE) 61.0%, Palmer (CHE) 54.0%, Konsa (ARS) 53.0%, De Cuyper (BHA) 38.5%, Hall (NEW) 32.0%.
Moved out: B.Fernandes (MUN) 50.0%, Mbeumo (MUN) 37.0%, Calvert-Lewin (LEE) 28.5%, Verbruggen (BHA) 25.0%, Maguire (MUN) 24.5%.
8.5% made no transfer and 7.5% took a points hit. Chips used, out of 200: wildcard 63, free hit 18, triple captain 11.
Source: data/fpl_elite_managers.json in the public repository, generated 2026-09-12.
Also free: your FPL career card on one shareable image, Saudi Pro League fantasy tools, and a creator program if you quote these numbers in your content.
Methodology
A Dixon-Coles style match model, tau corrected, fitted on recent results and the xG those matches produced. Clean sheet probability comes from the score matrix: the chance the opponent scores zero. Fixture difficulty is derived from win and clean sheet probabilities, ranked across every team fixture of the season and bucketed into five tiers. Fixture data comes from the official Premier League fantasy API.
More free FPL tools
Same model, same refresh schedule, no login.
- Captain picks: the pick and the contenders for the next deadline.
- Expected points: the top 100 of the projection, scoring rate and minutes apart.
- Model XI: the best 15 our numbers fit in 100.0m. Not the model's gameweek squad, which is frozen before each deadline.
- Predicted XI for every club: a projection, not a lineup leak.
- Differentials under 10% ownership.
- Price change watch: who moves tonight.
- xG, xA and xGI leaders.
- DefCon leaders under the current scoring rules.
- Points: projected vs actual, every player, every gameweek, with a permanent page per gameweek from gameweek 1 onwards.
- Player stats: shots, key passes, tackles, filterable, CSV.
- Defence profiles by pitch zone.
- Best player at every club, with the gap to the second option.
- Team news: everyone out or doubtful, by ownership.
- Notes from the model, only when the numbers say something.
- A page for every club: Arsenal, Liverpool, Bournemouth and the rest.
FAQ
- Which teams are most likely to keep a clean sheet in Gameweek 6?
- On GoalIQ's model the top clean sheet chances in Gameweek 6 of the 2026/27 Premier League season are Arsenal at 46.5% (home against Leeds); Newcastle United at 39.9% (away against Coventry); Hull at 38.6% (home against Everton). Projections update daily and sharpen as 2026/27 results accumulate.
- Is GoalIQ good for FPL?
- Yes. GoalIQ is built FPL-first: clean sheet probability, fixture difficulty, rate my team with a captain pick, a fit checker that searches for the strongest 15 around your must-have players, a draft rater (no team ID needed), price watch and the full xG/xA/xGI leaderboard for every player with data, a filterable player stats table with shots, shots in the box and key passes, are free, and GoalIQ Premium adds an interactive team manager with a gameweek planner, per-gameweek expected points (xP) for every player in the projection, the captain ranker, chip timing for the best Wildcard, Bench Boost, Triple Captain and Free Hit windows, transfer plans that chain 1 to 2 moves with hits priced in, an edge mode with rank-aware picks for your mini-league, a player value ranking, a DefCon tracker and transfer suggestions you can apply to your planned squad. Every number comes from a match model with a published, pre-match-logged track record.
- Is GoalIQ free?
- Yes. Clean sheet probability is free on the web and in the GoalIQ app for Android and iOS, and fixture difficulty is free on the web.
- How accurate is the GoalIQ model?
- The GoalIQ club model's predictions for domestic league and Champions League matches are logged before kick-off and never edited after it. Across the 553 completed club matches, it called the result correctly in 48.1% of them. In the 404 club matches that did not end in a draw, it called the result right 65.8% of the time (266 of 404). The model always names a side, so every draw counts as a miss and that 65.8% is the same 266 hits over a smaller number of matches. World Cup 2026 predictions came from a separate national-team model and have their own row in the prediction record. Every prediction is logged, hits and misses.
Disclaimer: GoalIQ provides model predictions and analytics. Not betting advice.