Gameweek 29

Hello everyone!! This is the second week of our Machine Learning Algorithm predictions and there are quite a few changes in the squad due to the double Gameweek (for ManCity and Arsenal).

For now, the coding part for 1 transfer per Gameweek is not integrated, so the algo can make unlimited transfers.

If you are here for the first time, please make sure to read the original post describing in detail the logic to predict top-performing players in the coming Gameweek and selecting the most optimal Fantasy team.

Previous Gameweek 28 Analysis

It’s time to check how our selected team faired in the Gameweek 28.

Let’s have a look at the scorecard:

• Algorithm: 35 Points
• Average: 26 Points

This was a low scoring Gameweek as ManCity, Sheff Utd and Arsenal did not have a match. Unfortunately, I have not coded the blank Gameweek module. So, algorithm 5 players from these teams. Even though, the algorithm did well compared to the average points.

The Gameweek 28 rank was 1,407,852 out of 7,450,900 which is around 18 percentile.

Predicted Points vs Actual Points

Well, there is a huge difference in predicted points and actual points.

Two matches resulted in unexpected wins,

• Watford (19th rank) 3 – 0 Liverpool (1st rank)
• Norwich City (20th rank) 1 – 0 Leicester City (3rd rank)

Liverpool unexpected defeat to Watford took algorithm and everyone by surprise and they are not invincibles anymore. A sigh of relief to Arsenal fans!! The dependable back 3 selected by the algorithm: Alisson, van Dijk and Gomez (didn’t play) collected a total of 2 points.

Other players also have very low scoring games, except Doherty and Deeney who surpassed the predictions by a margin.

• Doherty scored the opening goal against Tottenham and eventually, Wolves beat Tottenham by 3-2 in an exciting match, and
• Deeney scored and assisted against Liverpool

web_name	position	now_cost	predicted_points	actual_points
Alisson	GK	6.2	8.83701	1
Henderson	GK	5.2	7.305537	–
Lundstram	DEF	4.9	7.169078	–
Gomez	DEF	5.3	6.730101	–
Doherty	DEF	6.1	7.697514	12
Pereira	DEF	6.3	6.381388	2
van Dijk	DEF	6.6	8.265873	1
Jorginho	MID	4.9	5.325623	1
Fleck	MID	5.1	7.070911	–
De Bruyne	MID	10.7	7.586901	–
Richarlison	MID	8.3	8.377184	2
Snodgrass	MID	5.2	7.638534	1
Aubameyang	FWD	10.9	5.968566	–
Pukki	FWD	6.5	6.851487	2
Deeney	FWD	6.2	6.79892	11

Let’s predict top performing players and the best possible team combination based on the predictions.

Gameweek 29: Team Selected

As this is a Double Gameweek for Manchester City and Arsenal, the ML Algorithm has given more weightage to their players.

In the forward, Calvert-Lewin is introduced in place of Pukki as he has been quite dominating for the last few games.

In the midfield and defense, players from ManCity and Arsenal are introduced like Saka, Walker and David Luiz. Also, Ederson in the keeping position.

Gameweek 29: Points Prediction

The expected points of each player according to the algorithm is given below:

web_name	position	now_cost	predicted_points
Ederson	GK	6.0	6.843136
Krul	GK	4.5	5.013684
Pereira	DEF	6.3	6.843946
David Luiz	DEF	5.7	7.472146
Gomez	DEF	5.3	5.859144
Walker	DEF	5.7	7.410472
Doherty	DEF	6.1	7.447838
Richarlison	MID	8.3	8.191819
De Bruyne	MID	10.7	9.316884
Martial	MID	7.9	5.997385
Snodgrass	MID	5.2	4.995186
Saka	MID	4.6	7.380514
Calvert-Lewin	FWD	6.4	7.289779
Aubameyang	FWD	11.0	9.965698
Deeney	FWD	6.2	9.808444

Thanks for reading! See you in the next Gameweek.