Team #30680 Page 2 of 18
Contents
1 Problem Statement 3
2 Planned Approach 3
3 Assumptions 3
4 Data Sources and Collection 4
4.1 College Football . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4.2 Men’s College Basketball . . . . . . . . . . . . . . . . . . . . . . . 5
4.3 College Baseball . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
5 Network-based Model for Team Ranking 6
5.1 Building the Network . . . . . . . . . . . . . . . . . . . . . . . . 6
5.2 Analyzing the Network . . . . . . . . . . . . . . . . . . . . . . . . 6
5.2.1 Degree Centrality . . . . . . . . . . . . . . . . . . . . . . . 6
5.2.2 Betweenness and Closeness Centrality . . . . . . . . . . . 7
5.2.3 Eigenvector Centrality . . . . . . . . . . . . . . . . . . . . 8
6 Separating the Coach Effect 10
6.1 When is Coach Skill Important? . . . . . . . . . . . . . . . . . . 11
6.2 Margin of Win Probability . . . . . . . . . . . . . . . . . . . . . . 12
6.3 Optimizing the Probability Function . . . . . . . . . . . . . . . . 13
6.3.1 Genetic Algorithm . . . . . . . . . . . . . . . . . . . . . . 13
6.3.2 Nelder-Mead Method . . . . . . . . . . . . . . . . . . . . 14
6.3.3 Powell’s Method . . . . . . . . . . . . . . . . . . . . . . . 14
7 Ranking Coaches 15
7.1 Top Coaches of the Last 100 Years . . . . . . . . . . . . . . . . . 15
8 Testing our Model 15
8.1 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 15
8.2 Strengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
8.3 Weaknesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
9 Conclusions 17
2