Predicting Formula 1 Race Results with Machine Learning: A Deep Dive into Mariana Antaya’s Innovative Model
As the excitement builds for each Formula 1 grand prix weekend, fans often engage in lively discussions, speculating on which driver might claim victory. Data scientist Mariana Antaya has taken this enthusiasm a step further by developing a machine learning model that aims to forecast race outcomes. Her innovative approach has already seen success, accurately predicting the winners of three races in the current season.
Antaya, a self-proclaimed passionate follower of Formula 1, recognizes that machine learning is increasingly becoming a cornerstone in the world of racing, with teams leveraging various algorithms for real-time strategic decisions. "I think many fans aren’t aware of just how much teams utilize these advanced technologies," Antaya remarked during her conversation with motorsport enthusiasts. Her goal was to create a predictive model not only for fun but also to explore the potential of data available in the F1 domain.
To kick off her project, Antaya focused on lap times from the previous year’s Australian Grand Prix, which she obtained through the FastF1 API data repository. Her objective was to compare the anticipated results of the 2024 race with the qualifying performances recorded in 2025. By excluding rookie drivers from her model—due to insufficient data for accurate comparisons—she began the training process. Employing a gradient boosting algorithm, Antaya successfully predicted that Lando Norris would win the race at Albert Park.
Reflecting on her initial prediction, Antaya noted, "At the end of the video, I acknowledged that this was a relatively straightforward model, and I was pleasantly surprised by its accuracy." This initial success sparked greater interest within the F1 community, leading fans to rally around her project and suggest additional data points to incorporate into the model.
Antaya’s approach is highly collaborative. "I wanted to create a crowdsourced element in this project," she explained. "I encouraged the audience to suggest features they thought would enhance the model." This openness has allowed her to gather valuable insights on how to refine her predictions further, ensuring that the model evolves alongside the dynamic nature of Formula 1 racing.
As the season progressed, so did Antaya’s machine learning model. It continued to accurately forecast race winners, but she remained aware of its limitations. To enhance accuracy, she started incorporating more data points into her predictive framework. "The more data we have, the better the model can learn and adapt," she stated. "If we restrict the data, the model’s understanding is limited, which affects its predictive capabilities."
For the Japanese Grand Prix, Antaya introduced weather parameters, including the likelihood of rain and track temperatures at Suzuka. Additionally, she assessed drivers’ performances under wet conditions, which proved instrumental in her model’s successful prediction of Max Verstappen’s victory at that event.
The next significant leap for Antaya’s model came as she prepared for the Saudi Arabian Grand Prix. She integrated data reflecting each team’s performance throughout the current season. This additional layer of information enabled her model to understand shifts in team dynamics, such as McLaren and Williams making notable strides while others like Red Bull faced inconsistencies compared to their 2024 performance. "This comprehensive approach provides a clearer picture of team and car performance," she elaborated.
Antaya’s engaging content on platforms like Instagram and TikTok has gained traction, attracting attention from the broader F1 community. Some engineers from various teams have even reached out to her, expressing interest in her predictions. "The response has been astonishing," she admitted. "I’m genuinely surprised by how much interest there is. I have no concrete understanding of the sophisticated models employed by the teams, but I hope my approach is somewhat aligned with their strategies."
Despite her successes—three out of five race winners correctly predicted—Antaya is not resting on her achievements. As she looks forward to the Miami Grand Prix, she aims to experiment with more advanced machine learning techniques to further enhance her model’s accuracy and minimize the average error in her predictions. This average error serves as an indicator of the model’s performance, reflecting the difference between predicted outcomes and actual race results.
While aspirations for improved accuracy are high, Antaya recognizes that the unpredictable nature of Formula 1 presents inherent challenges. "There will always be variables that are difficult to quantify," she acknowledged. "For example, it’s challenging to predict the occurrence of a safety car during a race, which can trigger numerous cascading events."
Among potential enhancements, she mentioned the possibility of integrating historical data on crash rates during races, which could serve as another variable in the model. "While we can analyze past trends, the unpredictable essence of the sport means we can’t always foresee what will happen next," she said.
The journey of predicting Formula 1 race outcomes through machine learning is an evolving endeavor for Antaya. Her dedication to refining her model reflects the dynamic landscape of motorsport and the intersection of technology and sports. By harnessing data and engaging with the community, she embodies the spirit of innovation that defines modern Formula 1.
As the season unfolds, fans and analysts alike will be watching closely to see how Antaya’s model adapts and evolves, contributing to the broader conversation about the future of data analytics in motorsport. The blend of fan engagement, sophisticated algorithms, and a passion for racing may well reshape how we predict the thrilling outcomes of Formula 1 races.
