Users evaluate a player on a scale of 1-100 across five different categories (defense, passing, shooting, scoring, rebounding). Those values are then compared to percentile ranks of every NBA player by position that played over 250 minutes in the 2019-20 season (405 total players).
For example, if a user inputted the following: Defense: 10, Passing: 20, Shooting: 30, Scoring: 40, Rebounding: 50
The comparison algorithm would then go through all players by position selected (guard, wing big) and compare each NBA player’s percentile rank to the values that the user inputted.
Ex. Steven Adams comp score calculation for our given inputs:
-------------------------
We take this number (171) and subtract it from 500, then divide by 500 to get final comp score
We calculate this score for every player in the league and return the top ten scores. There is also an added coefficient (3) if there is a huge gap in any particular characteristic, to penalize players who are way off in one attribute. This creates a system where it is better to be close-ish across the board in all characteristics rather than close in many categories and really far off in others.