EPL Screaming Claret

This site is about the English Pemier League (EPL).

When games are played EPL Results are collected along with associated statistics, such as shots on target and expected goals. This data is assimilated into a model which updates the ratings of all teams involved.

Predictions for upcoming games are made, based on the updated ratings.    These predictions are in percentage terms for the game outcomes: Home win, Draw & Away Win (1X2).

The bookies' odds for these games are collected and compared with the predictions, in order to identify profitable bets. With a given bankroll, Kelly criterion is used to identify any profitable single bets. Generalized Kelly is used to see if backing multiple outcomes (e.g. away win and the draw for a given game) would be more favourable than any individual result.

The main prediction model is based on the assumption that goal differences in association football are logistically distributed. Then each team is given a rating, from which goal difference probabilities are calculated. Goal difference probabilities also imply the expected goal difference. When actual results are available the expected goals (xG) for each team is compared with the predicted goal difference, and the team ratings are adjusted accordingly.

Why use Expected Goals (xG)?

Expected Goals, or xG, are the number of goals that should have been scored when considering the number and type of chances in a match. It is a much more precise way of measuring performance than using actual goals scored. Better predictons of future performance can then me made, allowing profitable exploitation of bookies' odds.

about the author

footy engine