Model FAQ

Questions on this page:

  • What is the elevenify team ratings model?

  • How does the model update?

  • What are the model inputs?

  • How good is the model?

  • What do the numbers mean?

  • Is the model based on current season only?

  • What can I use the model for?

What is the elevenify team ratings model?

The model uses various inputs to generate baselines for a team’s attack and defence.

The model then uses these baselines to project goals for each match. These goal projections then go through a Poisson process to determine clean sheet probabilities and estimate the probability of match outcomes (win, draw, or loss).

How does the model update?

Short version: strength baselines undergo adjustments following each match. The adjustments are based on how the team’s performance aligns with the model expectations, taking into account match performance and the strength of the opponent. In other words, the model updates each week based on how wrong its predictions were.

Longer version: The model uses a ‘Bayesian’ approach. This means the model starts with placing great weights on its existing assumptions (based on historic evidence from the present back to prior seasons) and updates with new evidence as the season unfolds without completely discarding its existing assumptions. This update process involves adjusting the existing assumptions based on how well the new evidence aligns with or contradicts those assumptions. Importantly, the degree of adjustment is proportional to the strength of the new evidence. Strong and consistent evidence will lead to more substantial updates, while weak or conflicting information will result in smaller adjustments.

In my opinion, this approach is a great way to calibrate beliefs appropriately based on the strength of the evidence.

What are the model inputs?

Most models of this kind use a mixture of xG, penalties, finishing skill, and in the off-season blend in market spread data to capture those off-field elements that change over the summer.

My model primarily follows this approach with a couple of experimental differences.

How good is the model?

As a rough guide, this kind of model can generally be expected to have:

  • a season-long MAE of 0.9 to actual goals and 0.55 to xG.

  • a season-long RMSE of 1.15 to actual goals and 0.75 to xG

For reference, the projections are comparable to and can just beat out the popular but now-defunct 538 model.

The model performance has been documented and tracked by Ian on his blog.

What do the numbers mean?

Strength numbers will always indicate the model’s best estimate of strength at the time of update:

  • Attack Rating: predicted goals scored in a match vs average Premier League opponent.

  • Defence Rating: predicted goals conceded in a match vs average Premier League opponent.

  • Overall Rating: this is the predicted goal difference in a match vs average Premier League opponent.

Is the model based only on the current season?

No. The update method outlined in the ‘How does the model update?’ section above doesn’t ‘reset’ when a season ends - it is continuous and carries most of its beliefs across seasons with some adjustments.

What can I use the model for?

Whatever you like really! Most people are using the model to assist with Fantasy Premier League decisions whilst some others use it to help with betting.