Courtesy: Shutterstock

The unpredictability of football, no matter a team’s winning streak, is what riles the fans up to dawn their favorite team’s t-shirt, to howl and cheer for their team – till their voices run dry, to host “playful” animosity for the opposing team fans, to live vicariously through the winning goals and missed penalty shoot-outs. Despite the volatile nature of the game, fans have always tried to predict the winner of the matches.

Even before AI, Paul the Octopus was termed the “legendary soccer oracle” who took the world by storm, with its uncanny ability to forecast winners of the 2010 FIFA World Cup in South Africa. Throughout the tournament, Paul accurately predicted the outcomes of many matches, leaving fans awe-struck. Unlike AI, Paul’s predictions were shrouded with mysticism and the element of enigma surrounding football and its fans.

However, AI works in a far more systematic way using a wide range of data from the past. A large percentage of AI’s predictions for the 2023/24 Premier League season came to be true. AI Business partnered with Kickoff.ai, a startup born out of a research group at EPFL, to predict the 2023-24 season winners. By using sophisticated machine learning, they analyze vast amounts of data, including past team performance and player statistics, to identify trends and predict future outcomes. This can include a team’s tendency to score penalties or how their overall strength has changed over time.

Although Kickoff.ai failed to predict Aston Villa F.C.’s winning title, it was able to predict the season’s winner – Manchester City accurately. Kickoff.ai’s Lucas Maystre, also a research scientist at Spotify told AI Business, “It’s kind of expected that one or two such things are going to happen every season,” as according to him outliers will always appear.

The predicted probability of the Top 6 teams from last August (L) and the final top four (R) | AI Business/Kickoff.ai

AI’s Predictions for the Euro Cup 2024

Several AI analytics models have identified a clear favorite for the Euro championship. However, before we crown the alleged victor of this season, it’s crucial to assess the model’s accuracy by analyzing its performance in predicting past matches. Only then can we fully trust its prediction for the remaining games.

Opta’s supercomputer

Opta’s fabled supercomputer had predicted the performance of teams that entered the knockout stage. Before the tournament, England and France had an equal possibility of winning. But in reality, while England reached the finals, France lost its place to Spain in the semifinals. Similarly, Germany, another favorite according to the AI, lost to Spain in the quarterfinals. Spain has defied the odds of the AI predictions by reaching the finals, exceeding the probability that gave them a mere 9.6 percent chance of winning. However, Opta’s predictions also proved to be accurate on many occasions. The AI model predicted a close match between the Netherlands and Italy. This proved accurate in the quarterfinals, where the Netherlands emerged victorious from the group stage, while Italy fell to Switzerland and missed the semifinals altogether.

Courtesy: Opta Analyst

ChatGPT predictions

ESPN UK tapped into the ChatGPT crystal ball to predict the Euro 2024 winners, and their AI’s picks were strikingly aligned with those generated by Opta’s system. It is almost as though AI across is rooting for France to reach the finals; blindsided by Spain making it to the match held on 14th July 2024. As Lucan Maystre rightly said – there will always be outliers that the system can’t predict. In the case of Spain, the 16-year-old Lamine Yamal was that very outlier. He’s the youngest player to score at a men’s European Championship when he cemented Spain’s 2-1 semifinal win against 2018 World Cup winners, France.

Courtesy: ESPN UK, ChatGPT

Snowflake ML 
Snowflake ML also joined the fray of AI-powered Euro Cup prediction. They leveraged data from Kaggle (a machine learning and data community), incorporating historical information from the International Football Results database (1872-2024) and FIFA World Rankings (1992-2024). By meticulously structuring the data gathering, information coding, and model training process, Snowflake ML was able to generate predictions based on variables tied to this rich historical dataset. With a 65% accuracy rate just based on the group stage predictions, Snowflake ML proved to rank higher in terms of data analysis and prediction.

Predictions made by Snowflake ML

Courtesy- Snowflake Builders Blog.

Euro 2024: Actual Semifinal Winners

Courtesy- UEFA EURO 2024

The predicted winner of 2024 Euro Cup

England has been crowned the favorite by several AI models, with Kickoff.ai giving them a 19.2 percent chance of victory before the Euro Cup kicked off. But how do these predictions work, and should we put all our faith in them?

However, the story isn’t always set in stone. After the semifinals, ChatGPT, another AI model, re-evaluated its predictions based on player performance. Interestingly, it now predicts a Spain victory (2-1) with goals from Dani Olmo and Lamine Yamal for Spain and Harry Kane for England.

An industry that is benefiting in bounds, due to the integration of AI in sports is – the betting market. Bettors are reaping significant benefits from AI-powered tools, including increased prediction accuracy, advanced data analysis, and a personalized user experience. AI doesn’t just improve betting strategies, it also helps minimize potential losses. This is why platforms like BetIdeas and DeepBetting leverage AI to drive profits.

So, while AI can be a fascinating algorithm-based fortune teller, its predictions are as fluid as the game of football itself. New players emerge, and strategies shift; the only constant is the inevitable craze that erupts when a team wins.

• Share