The things people ask before they trust a number.
Straight answers about what the tool does, how we score ourselves, and what we do with your data.
Using the product
What do I actually get each week?
A starting eleven with a bench order, a captain and vice captain, the transfer that gains you the most, and a read on whether this is the week for a chip. Every one of those comes with the number behind it and a plain sentence saying why.
Do you connect to my FPL account?
No. We never ask for your FPL login and we do not sync your team automatically. You enter your squad, your bank and your free transfers yourself, and you copy any move you like back into the official site. Your FPL account stays entirely yours.
Is using this allowed?
Yes. FPL Master is an independent analysis tool. It reads public information, works out what it thinks, and shows you. You still make every decision and enter every move yourself on the official site. We are not affiliated with the Premier League or Fantasy Premier League.
How often do the numbers change?
The board refreshes through the week as team news, prices and availability move, and it settles into its final state before the deadline. If something big lands late, such as a manager ruling a player out in a press conference, the affected players update rather than the whole week being rebuilt.
What happens if a player is ruled out an hour before the deadline?
His chance of playing drops and everything downstream reacts: his projection falls, the lineup reshuffles, and if he was your captain you get told to move the armband. That is the moment the tool earns its keep.
What does it cost?
You can start free. The free account covers the player board, fixtures, news and our published accuracy record. The full weekly recommendation, the captain call and chip timing sit behind an account.
How we score ourselves
What does 80% to start actually mean?
It means exactly what it says. Take every player we have ever put at 80%, and about 80 in every 100 of them started. That property is called calibration, and it is the difference between a number you can bet on and a number that just sounds confident.
Why show a range instead of one number?
Because one number hides the thing you care about. Two forwards can both project six points while one is a steady six and the other is mostly two with a real chance of twenty. A captain choice is a question about the top of that range, not the middle of it.
What counts as a haul, and how do you get a chance of one?
A haul is a return big enough to change your rank, and we treat it as ten points or more in a single gameweek. The chance comes out of the simulation: we play the gameweek out many times and count the share of runs where the player clears that bar.
How do you decide whether you were right?
Three ways, and we publish all of them. Calibration asks whether our probabilities hold up. Points scored asks whether the squad we recommended beat the average manager. Points above template asks whether it beat the team the crowd was already on, which is the harder and more honest test.
Why not just publish one accuracy percentage?
Because a single percentage is easy to game and tells you almost nothing about a decision. A model can look accurate on average and still be wrong in exactly the spots where you had to choose. We would rather show the scorecard that can embarrass us.
What do you compare yourself against?
Two baselines. The average FPL manager, which tells you whether the tool is worth using at all, and the template team the crowd converges on, which tells you whether it is worth using instead of just following the herd. We report both, including the gameweeks we lost. The average manager figure is not our estimate: it is the number FPL itself publishes for every gameweek, and we check it against two independent public archives before we quote it.
Data and AI
Do you use AI, and what does it do?
Yes, in two clearly bounded places. Statistical models learn from ten seasons of match data to estimate how players are likely to return, and a language model turns those numbers into readable sentences and translates news headlines. The AI does not decide your team on a whim. Every recommendation traces back to a number you can see, and it never sees your password or your payment details.
What data do you learn from?
Publicly available Premier League and Fantasy Premier League match data going back ten seasons, together with published fixture, price and availability information. We do not learn from your personal data and we do not use your account to train anything.
What do you store about me?
The minimum to run an account: your email, your display name, your settings, and the squad you choose to enter. We do not sell it and we do not share it with advertisers. Operational records such as logs are cleared on a short cycle, and you can export or delete everything from Settings at any time.
Can I see the weeks you got it wrong?
Yes, and you should. The accuracy page shows every gameweek, including the ones where the average manager beat us. A tool that only shows you its wins is showing you a highlight reel.