Football Like Never Before
Has the data-driven epoch disenchanted football? Or is there a place for this relentless obsession with the quantifiable?
DISCLAIMER — It’s football, not soccer. And that is a fact, not an opinion.
In one corner, we have those who say that football has no place for statistics — that in football, moments that decide results are so few and far between that there is little use in trying to predict them.
In the other corner, we have those who argue that it’s all in the intricate details. It is about that extra goal that leads to those extra points and those extra points that lead to the extra trophies. One goal can be the difference that changes football forever.
As passionate football fans and analytics professionals, we believe we are uniquely positioned to present our piece on this heated debate. In our humble opinion, this is a false dichotomy. We believe football is a beautiful sport, and analytics is a unique lens to understand it better.
Resistance to change is normal. With increasingly rich and accessible footballing data, analysis using modern-day metrics has become ever-so-common. Even though analytics presents an enticing solution, one thing to always keep in mind is — Context Matters.
Football is one of the most fluid team sports there is, and momentum can change in seconds — the fans respond to a tackle, the atmosphere changes, and before you realize it, the game on the pitch has completely shifted. This shift in momentum cannot be captured or explained by statistics. But at the same time, the advent of statistics enables us to recognize opponents’ playing style patterns and identify weaknesses to exploit.
We can appreciate the art behind a Zinedine Zidane pirouette, but at the same time, also give Adama Traore credit for the most dribbles completed.
Off-the-ball movements that pull defenders off their position can contribute the most to the goal without even touching the ball. At the same time, unless you put the ball over the line and in the back of the net, no amount of movement, anticipation, and intelligence can win you the game.
Expect the Unexpected
Have you ever heard a commentator say something like — “He should’ve had a hattrick by now” or “He has missed a sitter of a chance.” These are precisely what statistics like Expected Goals (xG) can very easily explain.
The xG metric is the probability that a shot will result in a goal based on the characteristics of that shot and the events leading up to it. Some of the factors that determine this probability include the shot taker’s location, the shot’s body part, and the type of attack that led to the shot. When used correctly, such metrics can be very effective in understanding the sport. However, at the same time, at its heart, some of the methods may be considered reductionist.
With the 2022 FIFA World Cup fast approaching, we decided to take a step back and analyze the 2018 FIFA World Cup held in Russia. Using match event data and player performance data from ‘StatsBomb’ and ‘Fbref,’ we developed interactive dashboards in Tableau.
FIFA World Cup 2018 — Analysis
Fixture Analysis (click to access dashboard):
A detailed dashboard representing every match played at the 2018 FIFA World Cup. Along with basic summary stats of the game, a cumulative xG chart helps identify a timeline of goal-scoring opportunities, and the shot map provides a zonal analysis of shooting locations.
Player Analysis (click to access dashboard):
This dashboard facilitates analysis of a player’s performance throughout the tournament and during a specific match. Shooting locations can be used to assess the shot-taking ability of a player. In addition, pass maps and player heat maps present a good way to analyze a player’s overall performance.
Player Comparison (click to access dashboard):
This dashboard enables multi-stat comparison between players. The advanced spider radar chart gives a glimpse of different dimensions, whereas the quadrant compares one player to another. For this dashboard, we used advanced metrics such as npxG (non-penalty Expected Goals), successful pressures, shot-creating actions, interceptions, blocks, and aerial duels won percentage, among others.
xWorldCup — Statistics-modeled FIFA World Cup 2018
We used three key metrics to determine the outcome of every match in the tournament:
1. Probabilistic Performance Data — xG (Expected Goals) data for every match.
2. FIFA Country Rankings — The official rankings provided to all teams before the World Cup. It considers historical data about the match performance of the team during qualifiers and form leading up to the tournament.
3. Betting Odds — The betting odds generated before the World Cup represent the probability of a team winning the World Cup.
xGroupStage
With 32 teams from different continents competing for the highest honor in international football, this is how the xGroupStage turned out:
Uruguay, Spain, Brazil, Germany, and England all posted perfect records with three wins and a maximum accumulation of 9 points in the xWorldCup. However, in actuality, Germany (who were defending champions at the time) failed to qualify from their group in the 2018 FIFA World Cup. Here’s where the context comes into the picture. xG assumes average finishing ability for every shot. The same shooting opportunity provided to Cristiano Ronaldo has a much larger probability of converting into a goal than if provided to Phil Jones. For Germany, even though they racked up a large xG, their finishing ability let them down in their charge for the trophy.
The Uruguay team led by their captain and star defender Diego Godin had the lowest expected goals conceded in the group stage of 0.71. This means they conceded less than 1 goal across the 3 matches and were the best defense in the xGroupStage.
Iceland pipped Lionel Messi’s Argentina to qualify from the most closely contested group in the biggest shock of the group stage.
The cumulative xG of Group G was a massive 20.9! Group ‘G’ definitely stands for GOALS! With an xG of 7.28, Belgium was the best-attacking team in the group stage. They were followed closely by England, who had an xG of 6.5.
xKnockoutStage
Advancing into the tournament’s knockout phase, which teams were able to perform on the biggest stage?
Brazil went on to win what would be their 6th FIFA World Cup victory after beating Spain in the Finals by a delta of 0.35!
Portugal’s Trivela-maestro Ricardo Quaresma receives the Goal of the Tournament, scoring a chance of only 0.0123 xG — a 1.23% probability of the shot being a goal.
Honorable Awards
xGoldenBall — Brazil’s talisman Neymar Da Silva Santos Junior dominated the xWorldCup. With 4.8 key passes per 90, an expected assist every two matches, 8.4 shot-creating actions per 90, and 0.92 non-penalty goals per 90, he receives our xGoldenBall.
xGoldenBoot — Russia’s Denis Cheryshev scored 4 goals with a total xG of just 1.1. He was the player with the best finishing ability and receives our xGoldenBoot for this over-performance.
xGoldenGlove — Kasper Schmeichel’s post-shot xG conceded was 4.8; however, his goalkeeping prowess earns him the xGoldenGlove. He conceded only 2 goals and ended up with a save percentage of 94.7%, the highest in the tournament.
Co-Author — Shanay Shah (https://medium.com/@shanay.shah)
You can follow me and my work through these links — Website, Tableau, and Linkedin
