The world’s leading publication for data science, AI, and ML professionals.

EPL Game Week 6 Prediction using Data Science: xG Model

This is an article on my EPL Prediction series. You can check out the prediction for previous Game Week and how it held against the actual…

This is an article on my EPL Prediction series. You can check out the prediction for previous Game Week and how it held against the actual performance here.

Expected Goals or xG is the parameter used for prediction. If you are interested in understanding the algorithm for prediction, I recommend that you check out this article where it is explained in detail.

Analysis Up to Game Week 5

The above figures show xG Scored and xG conceded per match for each team. We can observe that the defending champions, Liverpool is head and shoulders above the rest in creativity. Spurs, Everton and Chelsea are also great attacking sides. The list also features the Villans and Hammers who are having a great run in the league so far. Westbrom lags behind in the list with around 0.5 expected goals created per match! Sheffield, Wolves and Crystal Palace also have less than one expected goals per match.

Aston Villa, the only team to have not dropped any points in the league, is the best defensive side with the lowest expected goals conceded. The Hammers, the Toffees, and the Seagulls also seem to have defence difficult to penetrate. On the other end of the spectrum, Manchester United and Westbrom have the most porous defence.

Based on xG Scored and xG Conceded, teams can be grouped into 4 quadrants as shown in the above graph. The horizontal dotted line shows the average xG scored per game. Teams above the horizontal dotted line are strong attacking sides and the teams below weak in attack. The vertical dotted line shows average xG conceded per game, teams to the left of have strong defence and the teams to the right have week defence.

The aim all teams should be in Q2 where both attack and defence are better than the average. For example, The Blades are one of the best defensive sides in the league. However they lack in creativity upfront. They Reds need to immediately address the problems in attack.

The Reds who are the best attacking side in the league are in Q1 as they are not very strong defensively. However their attacking ability is good enough to make up for the weakness in defence as indicated by the Delta xG Chart in the left.

A word of caution: All analysis is based on a very small sample size of just 4 or 5 matches per team and hence one bad game or good game tend to skew the conclusions.

Predictions for Game Week 6

Based on the performance up to Game Week 5, the predictions for Gameweek 6 are as below.

The absolute value of GD shows the competitiveness of the match. The higher the value, more one sided the match is expected to be and higher the accuracy of prediction. The lower the value of GD, the more the match could be anybody’s game. Positive value of GD means Home win and Negative value means Away team win.

Brighton Vs WBA is expected to be the most one-sided game with the Seagulls expected to win. Liverpool and Aston Villa also have high chance of winning their games. Spurs, Blues and the Toffees are also expected to return from their away fixtures with all the three points. Machester City, one of the favorites is expected to have a tough match against the Hammers, they must need some good luck not to drop points against a resurgent Moyes’ side.

Wolves vs New Castle is expected to be anybody’s game. Fulham at home against Crystal Palace and Leicester away at Arsenal have slight edge over their opponents.

The Red Devils hosting the Blues is expected to be the most high scoring match of the Game Week with the Blues expected to win. Despite the Blades being one of the best defensive sides in the league, their game at Anfield is expected to be a goal-fest. Fulham vs Crystal Palace on paper seems to be the most boring game of the weekend in terms of Goals.

Update: Actual Vs Prediction

Predicted Delta xG vs Actual Delta xG (Image by Author)
Predicted Delta xG vs Actual Delta xG (Image by Author)

Prediction for Arsenal Vs Leicester Match was the most accurate one whereas, the algorithm was least accurate in predicting Aston Villa vs Leeds Match (El Loco is a befitting title for Marcelo Bielsa, his tactics are crazy! to say the least. The algorithm is just a simple-baseline model, still I am hopeful that the prediction will improve with increase in sample size. Currently, the sample size is very small and predictions are hugely affected by outliers. Going forward, I am planning to factor in style of play, home advantage, finishing quality etc. to improve the model.

Please let me know your suggestions on how to improve the model.

Photo by Thomas Serer on Unsplash
Photo by Thomas Serer on Unsplash

Related Articles