Fantasy Basketball - Rank Predictor

May 20, 2016

182 views
109 downloads

Algorithms

Report Abuse
Creator - LawrenceB Summary - Who plays fantasy? If you do, you could utilize this model to help you predict next season player ranking.
Dataset - Yahoo 2015-2016 Season Stats + Rank Goal - Use current season stats and season ending ranking to predict next year ranking based on what you think the players will perform. Model - Boosted Decision Tree Regression Outcome - 88% Determined Coefficient Steps - 1. Enter projected stats for 2016-2017 or simply copy/paste this year stats if you think your players will perform the same. 2. The model will return "Scored Labels" = Rank (lower the better) ie PlayerABC & PlayerXYZ should be drafted higher than King James, if you think King James will produce the same stats next season as this season. The important thing to note is to enter next season projected stats. King James is rank 32.8 - due to lack of 3pt, low FT%, low 3pt%, high TO PlayerABC is rank 26 - better FT%, 2 3pt, high STL, low TO PlayerXYZ is rank 31 - better FT%, REB, 2 BLK, low TO NAME FGM FG% FTM FT% 3PTM 3PT% PTS REB AST ST BLK TO A/T Scored Labels L. James Cle 9.7 0.52 4.7 0.731 1.1 0.309 25.3 7.4 6.8 1.4 0.6 3.3 2.064 32.7920723 PlayerABC 5 0.45 2.7 0.82 2 0.35 15 5 1.8 1.8 0.5 1.2 1.5 26.3069973 PlayerXYZ 5 0.45 4 0.82 1 0.35 15 8 1.5 0.7 2 1.8 0.833333333 31.43296051 Notes - You don't have to enter all 15 stats categories. But the more you provide, the more accurate the model outcome will be.