Yet another basketball related project. Here's how my work fared in the 2018 season.
The left column shows the final predictions as of April 12, 2018. These predictions were made using the day-by-day model (see below)
The day-by-day model serves as a daily odds tracker for the MVP award.
Each day, mvp-predict would look to see which players come closest to the predicted statline or exceed it the most.
More stats met or exceeded result in a greater prediction percentage.
The forecasting model makes predictions prior to the start of the season.
The player that either meets or exceeds the most z-score benchmarks is the predicted MVP.
Since this requires foresight of how the top 20 players perform in 2019, I created a forecasting system
that projects statlines based on age and change in stats over the last two years.