Grading Our 2026 Oscars Machine Learning Predictions

Each year at BigML we use machine learning models trained on historical awards data to predict the winners of the Academy Awards. In our previous post, we presented our machine learning predictions for the 2026 Oscars (the 98th Academy Awards). Now that the ceremony has taken place, it’s time to evaluate how well the model performed.

The 98th Academy Awards ceremony took place on March 15, 2026, at the Dolby Theatre in Hollywood, honoring films released in 2025. One Battle After Another emerged as the big winner of the night with six Oscars, including Best Picture and Best Director.

Let’s review how our predictions compared with the actual results.

Prediction Results and Analysis

Across the main eight categories analyzed in our predictions, the model achieved:

• 7 correct predictions
• 1 incorrect prediction

This corresponds to an accuracy of 87.5%, which is a strong result given the inherent uncertainty of awards voting. Once again, machine learning proved capable of identifying the patterns and signals that often drive Oscar outcomes —from precursor awards to nominations momentum.

In addition to these predictions, we also predicted 11 more technical categories, of which we got 8 right. Below are the results:

As always, we encourage our users to explore their own ideas and experiments, such as incorporating additional data points that could further improve the public dataset. This approach aligns with BigML’s long-standing commitment to making Machine Learning accessible to everyone through transparent white-box models and workflows built on top of our proven algorithms. The Top Picks alone had an average 73% hit rate, whereas the coverage reaches 96% with the Top 3 taken into account.

Evolution of Our Prediction Success

The table below provides a cumulative view of our predictions from 2018 to 2026 and their performance in key Oscar categories. In addition to the standard Top Pick analysis shared in past posts, it also highlights how accuracy improves when factoring in the top two or three scoring nominees.

Final Thoughts

Predicting the Oscars is always challenging. Even with strong statistical signals from precursor awards and historical voting patterns, the Academy often delivers surprises.

Still, this year’s results demonstrate that machine learning can capture many of the patterns underlying Oscar outcomes. Our model successfully identified the winners in most major categories, achieving 89% accuracy.

At BigML, we are excited to continue this experiment next year by training new models, adding more signals, and exploring how machine learning can predict Hollywood’s biggest awards. For now, let’s enjoy all the movies that earned a nomination at the 2026 Oscars!