Creating Performance Prediction AI with Machine Learning

•

How I created an AI that can predict player performance:

Understand Machine Learning and Statistics. Understanding complex mathematical functions and theories helps me not only build the AI, but it truly gives me an added edge into finding top-notch value.

Obtain data. A lot of it. Data is the key, but simple statistics will not do. For example, the data I have just for MLB Player game logs came with 300+ different variables per player, per game. That ends up being tens of thousands of data points per projection.

Adjust. Adjust. Adjust. Making constant adjustments and improvements to the code not only is key in the beginning but after spending hours and hours looking at results from the AI, I can gain an understand at what the AI likes to do. For instance:

The AI for player points props in the NBA would often like to penalize stars who shoot a lot, but not efficiently. The issue with this? Those players shot a bad % but still went over their lines often enough due to crazy volume. Diving deep is often needed.

I could share more secrets, but maybe we save that for another day. Appreciate all the love you guys have given me this past year! Let's keep building something special!