Man City

Beta preditions.

Chelsea

Alpha predictions.

Liverpool

Alpha predictions.

Tottenham

Alpha Predictions.

Man United

Alpha predictions.

Leicester City

Coming soon.

How does it work?

Insider Knowledge

Lineup predictions are sourced from journalists, super-fans and other reliable sources.

Skill Ratings

Inputs are given a skill rating and greater weight is given to proven inputs.

Latest News

Predictions aggregate and interpret the latest news.

"Wisdom of the Crowd"

These features combine to produce lineup predictions with a likelihood percentage of start for each player.

Frequently asked questions

Each input has been given a ‘skill-rating’. This is a rating based on the accuracy of past predictions. Our lineup predictions are then produced with greater weight given to proven inputs which leads to more accurate results.

This is the percentage likelihood of a start created from adjusting the ‘raw’ predictions based on the input skill-level. Greater weight is placed on the predictions coming from inputs with a good skill rating and a lower weight is placed on those with a poor skill rating.

Please remember that these are ‘predictions’. The combination of inputs together creates a ‘consensus’ opinion on the likelihood of a player starting. For example, if a player is at 100% to start this does not mean they are ‘guaranteed’ to start, but rather, the inputs all have the consensus that the player will start.

The Man City predictions are our main predictions. Whilst they are still in Beta, they had an accuracy rating of roughly 80% during the 20/21 season. You can view the historic accuracy of the 21/22 season here.

The predictions are not always right, they are not immune to surprises and they are often wrong. This is a testament to how hard it is to predict Pep Guardiola. However, we do not think it is possible to have a better pre-match belief about the upcoming Man City lineup than these predictions.

All other predictions are currently in Alpha.

Please remember that these are ‘predictions’. The combination of inputs together creates a ‘consensus’ opinion on the likelihood of a player starting. For example, if a player is at 100% to start this does not mean they are ‘guaranteed’ to start, but rather, the inputs all have the consensus that the player will start.

This is the idea that combining multiple independent predictions will give more accurate predictions and help reduce ‘noise’.