How did AI Predict That Brazil Will Win The WorldCup?
The FIFA World Cup is one of the most widely celebrated football tournaments around the globe. The 2022 FIFA World Cup is expected to be watched by 5 billion viewers according to FIFA.
Enjoyed by the whole world, World Cup also attracts different traditions among the fans and supporters. Predictions are one of the most interesting traditions of the FIFA World cup. There have been numerous types of predictions that have been famous for different editions of the FIFA World Cup.
Paul the Octopus is one of the most famous predictors that rose to fame during the 2010 World Cup. By selecting between two clear boxes containing a treat, the octopus would choose between each box having the flag of the competing teams. Paul made eight predictions and each of them was correct.
Other than animals, psychics from around the world are also predicting different editions of the FIFA World Cup. However, in 2022, Artificial intelligence (AI) predictions have been in the limelight.
Different AI models have been trained for a couple of years to accurately predict the 2022 FIFA World Cup. A prediction model by Alan Turing Institute is one of the most famous AI predictions for the 2022 FIFA World Cup. So in this article, we will look into how did AI predict that Brazil wins the WorldCup and how accurate will the prediction might be.
The AI model from the Alan Turing Institute is a publicly accessible model that ran more than 100,000 simulations to predict Brazil as the winner of the 2022 FIFA World Cup.
The model was developed by Nick Barlow and his colleagues at the Alan Turing Institute. The Model uses data drawn from a compiled dataset by GitHub user martj42.
The dataset contains results and statistics of every international game since 1872. Then the researchers at the Alan Turing Institute fed all the data required since the 2002 World Cup into the model.
The data fed into the model also consisted of recent form and FIFA rankings of the world cup team. All the required data are passed through Bayesian Statistics. Based on Bayes’ theorem, Bayesian Statistics performs data analysis and parameter estimation to determine the probability of random events.
How did AI Predict That Brazil Will Win The WorldCup?
After applying this model to the 2022 FIFA World Cup, in 100,000 simulations, the AI predicted that Brazil has a 25% probability of winning the world cup. The AI predicted that the probability of Brazil winning the world cup is twice the probability of Argentina winning the world cup. Following is the table provided by the researchers at the Alan Turing Institute.
| Country | Round of 16 | Quarterfinals | Semifinals | Playoffs | Finals |
| Brazil | 92.3 | 74.4 | 53.4 | 53.4 | 25.1 |
| Belgium | 92.4 | 69.7 | 47.1 | 31.5 | 18.9 |
| Argentina | 90.7 | 61.4 | 43.3 | 23.5 | 13.2 |
| France | 87.4 | 60.0 | 40.6 | 21.7 | 11.0 |
| England | 80.7 | 58.5 | 31.7 | 15.5 | 7.1 |
| Spain | 76.8 | 42.4 | 20.5 | 10.7 | 4.8 |
| Netherlands | 86.5 | 54.8 | 26.3 | 11.1 | 4.7 |
| Denmark | 73.7 | 38.7 | 21.4 | 8.4 | 3.1 |
| Portugal | 80.6 | 38.7 | 17.2 | 8.1 | 3.0 |
| Croatia | 69.0 | 34.7 | 13.7 | 5.7 | 1.9 |
| Germany | 63.2 | 27.6 | 11.3 | 4.8 | 1.7 |
| Uruguay | 74.6 | 30.7 | 12.5 | 5.2 | 1.7 |
| Switzerland | 59.0 | 30.1 | 11.8 | 4.9 | 1.5 |
| Wales | 46.0 | 22.1 | 7.0 | 1.9 | 0.5 |
| Mexico | 45.6 | 14.1 | 5.3 | 1.3 | 0.3 |
| Poland | 46.7 | 14.4 | 5.4 | 1.3 | 0.3 |
| Serbia | 38.1 | 15.4 | 4.4 | 1.4 | 0.3 |
| Iran | 37.2 | 16.0 | 4.5 | 1.1 | 0.2 |
| Senegal | 50.5 | 17.4 | 4.6 | 1.0 | 0.2 |
| USA | 36.1 | 15.1 | 4.0 | 0.9 | 0.2 |
( Source: Newscientist.com)
As seeing the results of the group stage, Brazil is the favorite to win the world cup, and AI also backs it up. The second and third places were occupied by Belgium and Argentina respectively. Similarly, France finished in fourth place, while England was fifth.
Researchers of the model via press release reviled that they decided to use all international results from the 2002 FIFA World Cup onwards, after experimenting with their model’s success at predicting the 2014 and 2018 World Cups results.
However, the researchers at the university could not take weather, penalties, and individual brilliance into account. For the penalty shootout, the team used the simpler approach of assigning a 50/50 chance for which the team progresses rather than gathering data from the previous world cups.
There are other AI models like the model from Achim Zeileis at the University of Innsbruck in Austria, which also predicted a similar result. Their model also predicted Brazil as the winner with a 15% winning chance.
Read More: Rolls-Royce’s first ever electric car : Spectre – Specs, Design, & Price >>>
However, some models had different outcomes. Prediction model by the Insurance Company Lloyd predicted England will win the world cup by winning Brazil in the final. This same model had the correct prediction about the 2014 and 2018 FIFA World Cup winners. Furthermore, a model designed at the University of Oxford by Matthew Penn and his colleagues predicted Belgium as the winner of the 2022 FIFA World Cup.
