A Cost Efficient Approach to Building an NBA Team
Devised a ranking algorithm with self-defined KPIs and distance metrics to evaluate players’ performance and affordability, and accordingly built a "Fantasy Team".
View on GitHub
About the Research
While all players in the NBA are extremely talented athletes, many hone their skills to the point at which their expertise cannot be denied and simply demand massive salaries. This research presents a possible solution to the old of problem of creating the best possible performing sports team while at the same time minimizing the cost of the team.
Methodology
- Important Features: Field Goal Percentage (FGP), Defense Rebound (DREB),Offense Rebound (OREB), Rebound(REB), Assist (AST), Turn Over (TOV), Block (BLK), and Field Goal Made (FGM)
- Picked out most relevant features for each of the positions
- Ranked the selected Statistics within each position
- Calculated the weighted average of the ranks
- Created a plot for each position:
- x axis: standardized weighted rank; y axis: standardized salary
- Compared Euclidean Distances between each point and the origin
- Origin (Ideal player): free and has the smallest standardized weighted rank score (best performance)
- Selected the top 3 players for each of the positions
- Visualized sweet spots and unsweet spots for the NBA Finalists with Spatial Heat Maps and compared them with the team created by ranking algorithm
Further Details
For more information, check out the Research Paper here and the Presentation Deck here.