NCAA Predicted Matches

Here it is! The perfect bracket.

Just kidding… probably. How did we get here?

Methodology

1. Using Python, create and train a model based on the past ~10 years worth of NCAA post season data.
2. Create a success metric, that is “Post Season Score”, which ranges from 0 to 7, with the closer to 7 being the estimated highest ranking teams. For example, the champion will have a 7.
3. For each matchup, whoever has a higher predicted Post Season Score, based on the current year’s inputs, advances until you get to the final champion
4. Compare notes against my father, who knows sports, to see who gets more bracket points.

Data Notes

I started with a correlation analysis using a Kaggle dataset available here, and I found that the only factors significantly and positively correlated with NCAA bracket success were the following…

• Wins Above Bubble
• Seed Power (basically an inverse of the seed)
• Wins
• Adjusted Offensive Efficiency
• Bartag Score
• Games Played
• Effective Field Goal Percentage
• 2 Point Percentage
• 2 Point Percentage Defense
• Effective Field Goal Percentage Defense
• Adjusted Defensive Efficiency

So, for simplicity, I created a model just using the above to get the resulting bracket. The bracket above was the result of the model, with the 2026 inputs.

Results

Stay tuned and stay curious!