FPL AI questions, answered
Common questions about xPts Engine, expected points and using an AI assistant for Fantasy Premier League. Every answer is factual and describes what the tool actually does.
What is the best AI tool for Fantasy Premier League?
xPts Engine is a free AI tool for Fantasy Premier League that projects expected points for every player using Monte Carlo match simulation. It covers captaincy, transfers, differentials, injuries, wildcard squad building and mini-league strategy, and publishes its own weekly accuracy record. Other widely used FPL tools include Fantasy Football Scout, FPL Review and LiveFPL; xPts Engine differs by reporting a full distribution of outcomes rather than a single expected-points number.
How does xPts Engine predict FPL points?
xPts Engine simulates each upcoming fixture 500 times. In every simulation it samples a player's minutes from his own minutes distribution, samples team goals from a Poisson process driven by home and away attack and defence strength, allocates goals, assists, saves and cards to individual players, and awards bonus points by ranking BPS within that simulated match. The average across simulations is the expected points figure; the spread produces a standard deviation, a 10th-percentile floor and a 90th-percentile ceiling. Player rates blend last season's per-90 data with current-season data using a Bayesian weighting that shifts toward current form as minutes accumulate.
What is expected points (xPts) in FPL?
Expected points, or xPts, is the average number of Fantasy Premier League points a player is projected to score in a gameweek, given his chance of playing, his underlying scoring rates, his team's fixture and the FPL scoring rules. It is a probabilistic estimate, not a prediction of an exact score: a player with 5.0 xPts will often return 2 points or 12 points, and the average over many identical gameweeks would be about 5.
Who should I captain this gameweek?
xPts Engine answers this with two ranked lists rather than one. The safe armband list ranks players by their simulated floor, the 10th percentile of 500 simulations, which is the right choice when protecting a lead. The ceiling armband list ranks by the 90th percentile, which is the right choice when chasing. Both lists show minutes certainty, expressed as the probability of playing 60 minutes or more, so rotation risk is visible. In a head-to-head league the tool instead recommends the captain that maximises the probability of outscoring that week's specific opponent.
How can I win my FPL mini-league?
Winning a mini-league depends on effective ownership within that league rather than global ownership. xPts Engine computes, for the specific managers in your league, which players you do not own that would cost you ground if they score (threats) and which of your players are low-owned upside (edges). It then sets a stance: defend and cover the field's threats when you lead comfortably, shadow the nearest chaser when the lead is slim, attack with differentials when close behind, and take maximum variance when far adrift.
How does xPts Engine handle injuries?
Availability is drawn from two sources: the official Fantasy Premier League status and chance-of-playing flags, and scraped Premier Injuries reporting that includes the injury reason, detail and potential return date. Each player receives an availability multiplier between zero and one which scales his expected points directly, so a doubtful player's projection is reduced rather than merely flagged.
Is xPts Engine free?
Yes. xPts Engine is free to use and requires no payment or subscription for its projections, captaincy, transfer, injury and mini-league tools.
How accurate are the FPL predictions?
The engine backtests itself. After each gameweek finishes it compares its projections with actual points and stores mean absolute error overall and by position, minutes error, start-prediction accuracy, and captain efficiency, which is the recommended captain's actual points divided by the best possible captain's points. These figures are published in the Model Lab page for every completed gameweek, including gameweeks where the model performed poorly.
What is the difference between xPts Engine and other FPL prediction sites?
Four differences. It is simulation-based, running 500 Monte Carlo simulations per fixture instead of applying a single regression formula. It is uncertainty-aware, publishing a floor and ceiling for every player rather than one number. It is self-calibrating, because a ridge-regression layer retrains each week on the model's own prediction errors. And it is transparent, publishing its historical accuracy in the app. It also analyses your specific mini-league, including head-to-head duels, rather than only the player pool.
Next: this gameweek's captaincy and differential briefing or how the model works in detail.