FPL decision supportYou manage the squad.
We map the decisions.
Who starts, who gets the armband, and what an FPL transfer is actually worth. Every answer comes from Fantasy Premier League forecasts frozen at the deadline and scored in public, the hits and the misses alike. Pro is open to everyone until 30 September, no card needed.Start free, no card needed; Pro adds five-Gameweek planning and chip timing.
Running record: 2 scored Gameweeks, 1220 frozen forecasts, average miss 1.34 per player. Read live from the scorecard when you load this page.
Two scored Gameweeks is a small sample, and the caveat that came with the first one is still on the page. Read the full public scorecard →
Most FPL tools sell confidence.
We publish the size of our mistakes.
1.50 is the average distance between what SquadPilot said a player would score in Gameweek 1 and what he actually scored. Smaller is better. On its own it means nothing at all, so here it is next to the two free forecasts every FPL manager already has, scored on the same 600 players in the same Gameweek.
Small edges are the only kind that exist.
SquadPilot's typical miss was a tenth of a point smaller than the estimate FPL prints for free. That sounds like nothing. It is applied to fifteen players a week, to a captain who counts double, across thirty-eight Gameweeks. That is the whole business. Nobody in fantasy football wins by a mile. They win by a tenth, repeatedly.
The chart above starts at zero and stops at two, because that is the real scale. We could have cropped the axis and made that gap look like a canyon. A company selling calibration does not get to cheat its own chart.
And here is why that result is weaker than it looks.
In Gameweek 1 FPL has no current-season data, so it hands the same estimate to whole groups of players at once. 575 of the 600 sat on one of those repeated default values. A number handed out in bulk is a placeholder, not a forecast, and beating it proves very little.
On the 25 players FPL actually priced individually, FPL's estimate beat ours by 0.20. That is the honest version, and it is on the scorecard in the same table as the good number. One Gameweek proves very little on its own.
De Cuyper: we said 2.3. He scored 17.
That is the single worst call SquadPilot made in Gameweek 1, and it is named at the top of the scorecard rather than buried under an average. Every points model has a De Cuyper week. The ones that don't show you theirs are choosing not to.
Frozen forecasts are written once and never rewritten. There is no path in the product that edits a call after the whistle, which is exactly what makes the good numbers worth reading.
Averages hide the part that decides your rank.
Broken out by what each player really did in Gameweek 1, the average miss was 0.66 on players who didn't play, 1.20 on blanks, 3.21 on returns of 3 to 9, and 9.49 on hauls of 10 or more.
Hauls are where every points model in the world hurts, and where this one has the most to prove. That is the honest starting line, published before anyone asked for it. See the full scorecard →
The four questions every deadline asks.
Answered with numbers, not vibes.
Expected minutes that read the team news
Start probability on every player, built from pressers, flags and rotation patterns, refreshed every few minutes as the deadline closes in. Every number carries its reasoning.
The armband call, with the spread shown
Captains ranked by projected points, and for each pick the floor, the ceiling and the odds of a 10+ haul. A safe 8 and a coin-flip 14 are different bets. You see which is which.
The swap priced over five Gameweeks
What the move is worth across the run of fixtures ahead, with bank, hit maths and price included. And when the honest answer is no move at all, it says so.
Both chip sets, planned against the fixtures
Two chip sets run this season, and the first expires at the halfway deadline. The chip planner times Wildcard, Bench Boost, Triple Captain and Free Hit across both, before the panic week arrives.
A full argument, not a tip.
Every recommendation arrives in four parts: the action the data supports, the expected return behind it, the risk it can see, and the alternative if you read the game differently.
Below is the shape every recommendation takes. Your own is built from live FPL data against your actual squad, and it names real players and real numbers. You are the final judge. SquadPilot recommends, you make the moves.
Free gives you the weekly squad check: your captain, a suggested XI from your own squad and which slots need attention. Pro gives you the full recommendation: the exact personalised transfer, whether to roll or move, the alternatives and how to fund them.
Frozen at the deadline.
Scored in public.
No hindsight edits, no quiet deletions. Every projection is written to the Decision Ledger: append-only, frozen and timestamped, then graded against what actually happened. Including the reads it gets wrong.
Inputs refreshed 30 min before. Frozen, hashed and timestamped. No takebacks.
The pitch doesn't care what anyone projected.
Hits and misses, side by side, permanently.
The live match centre: the forecast takes its medicine in real time.
During every fixture, actual points stream in next to the frozen forecast, refreshed about every two minutes. Each player moves through In play → Awaiting confirmation → Final, so you watch the calls land or miss live, not in a write-up afterwards. The prediction stays on screen while reality grades it.
Frozen at the deadline
Inputs refreshed 30 minutes before the deadline, then the prediction is frozen and timestamped. What it said is what it said.
Misses stay on the record
When the model reads a fixture wrong, the ledger says so, openly, next to the score. Trust built on hindsight edits isn't trust.
Your overrides are scored too
Ignore a recommendation? The path you chose and the path it suggested are both graded. The ledger cuts both ways.
What landed after the first two Gameweeks were graded.
A scorecard is only worth reading if the product behind it keeps moving. Everything below is running now, not on a roadmap.
Who is actually on penalties
Penalty, corner and free-kick order for all twenty clubs, in the order the takers stand in, and next to each name, how long that duty has been held. A taker promoted last week and one who has had the ball for a season are not the same bet, and the date says which is which.
PER CLUB · ORDER + HOW LONG HELDA predicted eleven for every club
A predicted eleven for every club, assembled from the start probabilities the minutes model already produces and kept to a legal formation. Each carries a confidence read off the tightest within-position margin, and every contested place is named: who is in, who is pushing, and by how much.
20 CLUBS · CONFIDENCE + CONTESTED PLACESTeam-news sources get scored, not trusted
Lineup sites publish accuracy figures about themselves. SquadPilot scores their claims against the line-ups that actually took the field instead. Before every deadline the engine files its own predicted XIs under the source name squadpilot-engine and is graded by that same code. We do not get to mark ourselves by a kinder rule than the sources we mark.
GW1 · FIRST SOURCE SCORED: 220 CLAIMS, 80.45%Our own team, on the record
SquadPilot's own team, run under the full FPL rules and scored with FPL's official points: one budget, the same transfer limits, the same deadline. It is not an entry in the official game. It is picked from the model's own numbers, published, and frozen each deadline. The scored history sits on the public record as it accrues; where nothing has been scored yet, it says so rather than showing a flattering blank.
FULL FPL RULES · OFFICIAL POINTS · FROZEN EACH DEADLINE SEASON SO FAR · … ·Where you rank, and where you differ
Overall rank, your country's table and every mini-league you're in, with the direction you moved. Then the part that decides those leagues: who your rivals own, who they captained, the effective ownership behind it, and exactly where your fifteen departs from theirs. A rank FPL doesn't know is left blank, not shown as a zero.
RIVAL OWNERSHIP · CAPTAINCY · EFFECTIVE OWNERSHIPEvery transfer you made, marked
Each transfer is scored on what the two players actually did next, the one who left and the one who arrived, with any points hit charged against it. The ledger already grades the model. This is the half that grades the manager.
OUT vs IN · HIT CHARGEDThe live match centre
While a ball is in play, actual points stream in beside the frozen forecast about every two minutes. Bonus is derived from live BPS before FPL awards it, and marked as an estimate rather than a fact until it is confirmed. Defensive contribution shows the actions so far against the threshold for that position.
REFRESH WHILE IN PLAY · ~2 MINUTESFixtures priced, not just tinted
Every fixture in the horizon carries expected goals for, expected goals against and a clean-sheet probability, the same numbers the player projections use. FPL's own difficulty rating is passed through untouched beside them and kept visually separate, so a model figure is never mistaken for an official one.
PER FIXTURE · xG FOR / xG AGAINST / CS%Every pick shows its spread
Every player carries a floor, a ceiling and the odds of a 10+ haul, never a bare number. A dependable six and a coin-flip fourteen are completely different bets, and the armband decision is usually a choice between exactly those two.
SHOWN PER PLAYER · FLOOR / CEILING / HAUL ODDSIt knows what you've already spent
Chips you have played are read straight from your FPL history, so the planner never suggests a Wildcard you burned in August. Two chip sets run this season and the first expires at the halfway deadline, and both are planned against the fixtures.
READ FROM YOUR FPL HISTORYExpected minutes that hear the press conference
Start probability tightens as team news lands. Inside the final hour before a deadline the read refreshes every five minutes, and each estimate carries a freshness stamp, so you can tell a fresh observation from a historical prior at a glance.
FINAL HOUR REFRESH · ~5 MINUTESInstallable, and honest when it's offline
Add it to your home screen and the shell is cached, so it opens on a bad signal. The API is never cached. A projection, a price flag or a deadline served from yesterday would look exactly like a fresh one while being quietly wrong, so when the network is gone it says so instead.
APP SHELL CACHED · API NEVER CACHEDHow the FPL planner works: five steps, squad to scorecard.
- Import your squadJust your team ID: public data, never your password.
- SquadPilot analyses your positionFixtures, form, prices and flags, read against your squad.
- Compare the pathsAction, expected, risk and alternative, side by side.
- Freeze your decisionYour call goes on the record before the deadline.
- See how it scoredThe ledger grades every call after the final whistle.
Want the method rather than the summary? Read how the projections are built →
Free to start. Founding prices while they last.Free to start. Pro when you want the depth.
| Free | FoundingPro | |
|---|---|---|
| Model runs | 3 a day | Unlimited |
| Weekly squad check: captain, suggested XI, slots needing attention | Included | Included |
| The exact personalised transfer, roll or move | Not included | Included |
| Alternatives with funding routes | Not included | Included |
| Check your own transfer ideas | Not included | Included |
| Recalculate with your own xMins | Not included | Included |
| Weekly Picks, Free Hit and Wildcard teams | One featured pick | Full |
| Decision Ledger | Full access | Full access |
| Planning horizon | This Gameweek | Five Gameweeks ahead |
| Chip planner, both chip sets | Not included | Included |
| Risk profiles (conservative, aggressive) | Balanced only | All three |
| Hindsight: every transfer you made, marked | Summary | Full |
| Mini-league rivals | Summary | Full |
| Saved plans | No | Up to 10 |
Pro is open to everyone until 30 September. That is the Pro Preview. The Pro Preview ended on 30 September. Anything you built during it is kept, just locked: you can still read it, and Pro is only needed to recalculate or edit it. Founding launch pricing runs until 30 September 2026. From 1 October, new purchases move to £29.99 for the Season Pass or £3.99 for a 30-Day Pro Pass.Pro is £29.99 for the 2026/27 season (until 31 July 2027), or £3.99 for a 30-Day Pro Pass: one payment, no auto-renew, renew when you want.a month for Pro Monthly, which renews each month until you cancel.
Enough to see exactly how the model thinks. The cap is on how often you run it, not on what it shows you.
- Pro open to everyone until 30 September: the Pro Preview, no card neededWork from the Pro Preview is kept, just locked until you upgrade
- 3 full model runs a day
- Your weekly squad check: captain, suggested XI from your own squad and which slots need attention
- Full Decision Ledger access
- The public Scorecard
One payment covers the 2026/27 season (until 31 July 2027). Every feature, every Gameweek.
- The exact personalised transfer and the roll-or-move analysis
- Alternatives with funding routes, and checks on your own transfer ideas
- Recalculate with your own xMins
- Full Weekly Picks, Free Hit and Wildcard teams
- Unlimited model runs, all season
- Five-Gameweek planning & chip timing
- Conservative and aggressive risk profiles
- Full Hindsight and mini-league rivals
- Up to 10 saved plans
Every Pro feature for 30 days. One payment, no auto-renew.Every Pro feature, month by month.
- Everything in the Season Pass
- 30 days of Pro from the day you payRenews each month until you cancel
- Nothing renews: buy another pass when you want oneCancel any time; Pro runs to the end of the month you paid for
- Bought the monthly product before 1 October? Each new pass is still £2.99 for you until the end of the 2026/27 season
Free stays free either way. The founding prices end 30 September.Founding prices ended on 30 September.
Payments are live via Paystack and charged in South African Rand: about R460R691 for the Season Pass, R70R94 for the 30-Day Passa month, at current rates.
Straight answers.
Is SquadPilot affiliated with FPL or the Premier League?
No. SquadPilot FPL is fully independent. It reads only public data from the official Fantasy Premier League API. It never asks for your FPL password and never stores one.
What's actually free?
Until 30 September, everything: the Pro Preview gives every account full Pro depth free until the end of that day, no card required. After that, the free tier keepsThe free tier keeps three full model runs a day and your weekly squad check: captain, suggested XI from your own squad and which slots need attention. It also keeps full Decision Ledger access and the public scorecard. That tier is a working product and it stays free. Pro adds the exact personalised transfer, the roll-or-move analysis, alternatives with funding routes, checking your own transfer ideas, xMins recalculation, five-Gameweek planning, chips, full Weekly Picks, Free Hit and Wildcard teams and saved plans.
What is the Pro Preview?
Until 30 September 2026, every account, free accounts included, gets the full Pro feature set: the five-Gameweek planning horizon, chip planning across both chip sets, the conservative and aggressive risk profiles, up to 10 saved plans, and Hindsight and mini-league rivals in full. No card required. It exists so you can judge the product on its output before spending anything.Until 30 September 2026, every account, free accounts included, had the full Pro feature set, with no card required. It has now ended. Anything you built during it, saved plans, chip plans and scenarios, is kept, just locked: you can still read it, and Pro is only needed to recalculate or edit it.
When does billing start?
Only when you buy Pro, and only once per purchase. Pro is £29.99 for the Season Pass, one payment for the 2026/27 season (until 31 July 2027), or £3.99 for a 30-Day Pro Pass. The pass is one payment for 30 days of Pro with no auto-renew: nothing is charged again, and you renew when you want. Founding Launch pricing ended on 30 September 2026. If you bought the monthly product at the founding price before then, every 30-Day Pro Pass you buy is still £2.99 until the end of the 2026/27 season (31 July 2027). Payments are processed by Paystack and charged in South African Rand. The free version remains free.
Can it make transfers on my FPL account?
No. It can't and it never will. SquadPilot recommends. You decide, and you execute on FPL yourself. It has no access to your account and no way to act on it.
Does SquadPilot run its own team?
Yes, and it is a simulation, not an entry in the official game. The House Team is SquadPilot's own team, run under the full FPL rules: one budget, real free transfers and selling prices, hits charged, chips under FPL's chip rules and the same deadline. It is scored with FPL's official points. The squad is picked from the model's own numbers, published, and frozen at each deadline like any other forecast. Its scored history goes on the public record as Gameweeks land on it, and where nothing has been scored yet the page says so rather than showing a flattering blank.
Can I install SquadPilot on my phone?
Yes. Add it to your home screen from your browser and it installs as an app. The shell is cached so it opens on a bad signal, but the API is never cached. A projection, a price flag or a deadline served from yesterday would look identical to a fresh one while being quietly wrong, so when the network is gone the app says so instead of showing stale numbers.
Is SquadPilot more accurate than FPL's own projections?
On two scored Gameweeks, on the full population, yes, though the caveats travel with the headline. In GW1 the average error was 1.50 points per player against 1.60 for FPL's own free estimate, on the same 600 players; 575 of those 600 sat on a repeated default value FPL hands out in bulk when it has no current-season data, and on the 25 players FPL priced individually FPL's estimate beat ours by 0.20. In GW2 the frozen forecast came in at 1.19 against 1.46 for the same FPL estimate, on 620 players, and on the 29 individually priced players we were ahead by 0.08. Every one of those numbers is on the scorecard. Two Gameweeks is still a measurement, not a trend.
What happens when the model is wrong?
It stays on the record. In GW1 the model forecast 2.3 points for De Cuyper. He scored 17. In GW2 it forecast 6.2 for Bruno Fernandes, who scored 23. Both misses sit on the public scorecard permanently, next to the hits. A model you can't audit is a model you shouldn't trust.
Who builds SquadPilot.
SquadPilot FPL is built and run by Nashlyn Pather and Darrshan Govender. It is independent of the Premier League and Fantasy Premier League, reads only public FPL data, and puts every forecast it makes on the public record. Questions, corrections and bug reports go to support@squadpilot.tech.
Every Gameweek locks. The projection freezes with it.
Pro is open to everyone until 30 September.Start free, no card needed. Judge it on the scorecard, not the sales pitch.
Start free