Analyzing Chelsea Signings for 2020–21 Premier league with Statsbomb inspired Radarplots

Premier League giants Chelsea were in most talked about team during the Summer transfers. The six-time English champions from London splashed €247.2 million ($292.4 million) to acquire German players Kai Havertz and Timo Werner, English defender Ben Chilwell, Moroccan winger Hakim Ziyech, Senegalese goalkeeper Edouard Mendy, also signed experienced Brazilian center back Thiago Silva and French defender Malang Sarr on free transfers.¹

The aim is to analyze the performances of these high-profile signings, with their previous year performances. Since the 2020–21 season is yet to be finished I have taken data for the 29 matches which have happened so far. The idea is the create a radarplots which are inspired by Statsbomb who are leading football analytics experts. And I will be taking help from open-source library for python which is soccerplots to visualize and interpret the results.

Data gathering

The most laborious task of all is gathering accurate and relevant data for visualizing. Thanks to FBrefs, a website that gives overall statistics of a match, including players' performances on the pitch. The data includes:

1. Passing and Pass Types
2. Goal and Shot Creation Stats
3. Defensive Actions
4. Possession Stats
5. Miscellaneous Stats (includes Aerial won).

this data is readily available on the website and one can export these statistics into an excel file for exploration.

I have chosen different Performance Indicator for different positions from the data. For example: for Center back (defensive player)we will be looking at:

Total Passing Completion% (TotalCmp%), xG Assisted (xA), Progressive Passes (Prog), Fouls per game (Fls), Aerials Won% (Won%), Aerials Won (Won), Pressures% (% of time team gained possession by apply pressure), Tackles/game (Tkl), Interception/game (Int), Clearance/game (Clr), Blocks/game (Blocks), Error led to goal (Err).

Comparing Thiago Silva’s performances with PSG(19–2020) and Chelsea(20–2021)

After superimposing performances from both seasons, it becomes easier to compare. Error led to goal(Total Error led to goal / Total number of matches played) was greater for the 2019–20 season with PSG i.e. 0.09 when compared to 0.06 for Chelsea this season. Also, Aerial won% is higher with Chelsea. Of course, we cannot discount that data belongs to 2 different leagues with different playing styles. But one thing is sure, Thiago Silva has

Comparing Ben Chillwell’s performances with Leicester City(19–2020) and Chelsea(20–2021)

Chillwell will not be happy watching this, clearly % Dribblers tackled and xA were significantly higher the previous season. But areas such as Progressive Passes, Interception/game, and Pressure% are improved.

Of course, not all factors which go into player performances on the field are quantifiable, but such a way of visualization of data helps in getting a good picture of how they’re performing.

[1]: https://www.aa.com.tr/en/sports/englands-chelsea-spend-over-292m-in-summer-signings/2002716#:~:text=Chelsea%20spend%20nearly%20quarter%20billion,Chilwell%2C%20Ziyech%2C%20Mendy%20in%202020&text=English%20Premier%20League%20club%20Chelsea,stellar%20players%20during%20the%20summer.

This article’s Python notebook: Available on my GitHub Repo.