Before even hitting a ball, the Opta supercomputer had already played the tournament 25,000 times. Her verdict: Spain was the most likely winner, winning 16.1 percent of the simulations, followed closely by semi-finalists France, England and Argentina, each winning more than 10 percent.
The forecast quickly became much more than a percentage on a page. It graced television screens, influenced the betting markets, dominated discussions on social media and gave millions of fans a new way to follow the tournament. What started as a prediction experience became part of how the World Cup was experienced in real time.
How the model works
As Jonathan Whitmore, director of analytics at Stats Perform, explains, the model combines the Opta team’s assessment with betting market odds. Team ratings are built on an Elo system, the same family of models used by FIFA. Elo weighs not only wins and losses, but the actions behind them. “Each respective team has a score, and if you beat a higher-scoring team, you have the ability to earn more points and increase your score, while if you lose against a weaker team, they effectively take those points from you, so it adjusts over time,” explains Whitmore.
Germany’s shock penalty shootout loss to Paraguay in the round of 32 illustrates how the system adapts. The upset doesn’t just cost Germany a place in the draw, it actively transfers rating points to Paraguay, reshaping both teams’ chances in each subsequent simulation.
However, ratings alone cannot capture everything that shapes a match, so the model relies on a second source of data to fill that gap: the betting markets. “We indirectly calculate information such as injuries, team selection from betting odds,” says Whitmore, “so we are able to more accurately predict those future games coming up over the next few weeks.”
Before the tournament, the model ran 25,000 simulations, significantly more than the 10,000 typically used for competitions such as the Premier League due to the smaller number of total World Cup matches. That degree paid off. At the semi-final stage, Spain, France, England and Argentina, the model’s top four pre-tournament teams, were also the tournament’s final four.
The value of AI predictions
Prediction has evolved from a pre-match novelty to one of sports’ most valuable products. What was once a tournament preview is now a live mirror feed that powers broadcasts, enriching betting experiences and keeping fans engaged from the opening whistle to the final. Built on the trusted Opta data that underpins modern football, the Supercomputer has become one of the sport’s most trusted benchmarks.
The model is constantly updated. “For every goal, every red card, every full-time whistle, every penalty, we get an updated simulation showing who is most likely to win the tournament,” explains Whitmore.
For example, before the semifinals, France had overtaken Spain as the tournament favorite, winning 34 percent of the model’s simulations. This constant recalculation is what makes modern forecasting so valuable. The tour becomes a live narrative, with every moment reflected in an ever-evolving forecast.
This reflects a wider shift in sports media. As data becomes more sophisticated, it has also become part of the story. Each goal changes the probability of lifting the trophy. Each red card reshapes a team’s path to the final. Any result elsewhere recalculates the chances of qualification. Prediction models deepen the drama by giving each key moment additional context, revealing how the match and the tournament evolve with each new development.
Watch any major soccer broadcast today, and these insights are everywhere. The odds of winning decrease near the score line. The graphics of the moment ebb and flow with the game. Qualification scenarios are updated instantly. These visuals no longer feel new because they have become part of how football is viewed and understood.
This represents a fundamental change in fan behavior. A decade ago, expected goals were a specialized metric debated by analysts. Today, it is part of the everyday football conversation. Prediction models are following a similar trajectory, moving from niche analytics to the common language of sports.
For broadcasters, this context creates ongoing storytelling opportunities. Each probability update creates another talking point, another graphic, another social clip, and another reason for viewers to stay engaged. Together, they turn a football match into an evolving narrative, with live prediction models that update during the game.
BBC Sport demonstrated this during Scotland’s crucial Group C match. Using Opta’s live prediction data, viewers can watch Scotland’s chances of reaching the round of 32 throughout the match. The model predicted Scotland would advance if they lost by no more than two goals, raising the stakes with each attack. Brazil’s goal eventually erased that margin, ending Scotland’s World Cup run in real time.
This is why fans appreciate the prediction. The appeal is not the percentage itself, but the context it provides. Every goal, save and red card creates an immediate change in the narrative of the tournament, giving fans another reason to celebrate, argue or fear. with 93 percent of General Z by using a second screen while watching sports, live previews naturally extend the experience beyond television, encouraging audiences to follow the story across multiple platforms.
The same data can also power personalized experiences. A casual fan may simply want to know who is most likely to win the tournament, while a dedicated supporter wants to understand their club’s or country’s changing path to the final. A trusted predictive model can simultaneously support stream graphics, editorial coverage, fan experiences and betting products.
In this way, predictive models are becoming the connective tissue between live sports data and audience understanding. As the volume of information grows, from historical performance to player tracking to real-time match events, the challenge is to understand what the data means in the moment. Predictive models provide that missing layer of interpretation, transforming raw inputs into narratives, recommendations and decisions across streams, personalized fan experiences, betting platforms, notifications and new AI assistants. With sports data becoming more abundant and complex, predictive models provide the context that turns information into meaning. Companies that control this explanatory layer will shape the way audiences experience live sports.
Trust as the ultimate competitive advantage
For sportsbooks, broadcasters and media companies, the commercial value extends even further. Prediction models help explain how a match is evolving before and during play, giving streams, betting products and digital experiences a common layer of trusted context. Live probabilities, predicted lineups and tournament predictions increasingly support products and services built around live sports. None of this works without trust.
Broadcasters will not build programs around predictions they do not believe. Sportsbooks will not integrate unreliable models into customer experiences. Fans will not return to predictions that consistently fail to reflect the play on the field.
This is why accuracy carries so much weight. A wrong end result is quickly forgotten. A repeated misprediction in streaming graphics, betting products, editorial content and push notifications becomes a credibility problem multiplied on any platform that relies on it. As prediction becomes more valuable, trust becomes even more important.
Perhaps the clearest demonstration of that belief came not from a broadcaster or a bookie, but from the man who runs world football himself. Asked about installing Spain as pre-tournament favorites by Supercomputer Opta, FIFA President Gianni Infantino just smiled and replied, “Well, if Opta says so.”
It was a light-hearted remark, but it captured something significant. The Opta supercomputer and prediction models are trusted not because they use AI, but because they are built on the data that millions of broadcasters, sportsbooks, clubs and fans already rely on to understand the game.
As AI continues to evolve, predictive models will become even more sophisticated. However, their greatest value will remain the same: helping millions of people understand, in real time, how each moment changes the story unfolding before them.
