Tuesday, March 9, 2010

Tourney Simulator

I created a tourney simulator a year ago that took a basic formula to determine the winner of each game.

The seeding for the individual team will determine how many randomly-generated "draws" it needs to have before they win. For instance, Team 1 is seeded 1, and Team 2 is seeded 16, therefore a randomly generated draw of either 0 for team 1 and 1 for team 2 would indicate a "win". Team 1 would need one win. Team 2 would need 16 to win the matchup. I did this simulation for all first-round matchups (8 games, 16 teams) 100, 1000, and 10,000 times:
1 v. 16
Team 1 -> 100 100.0%
Total Records: 100

2 v. 15
Team 1 -> 100 100.0%
Total Records: 100

3 v. 14
Team 1 -> 100 100.0%
Total Records: 100

4 v. 13
Team 1 -> 99 99.0%
Team 2 -> 1 1.0%
Total Records: 100

5 v. 12
Team 1 -> 93 93.0%
Team 2 -> 7 7.0%
Total Records: 100

6 v. 11
Team 1 -> 93 93.0%
Team 2 -> 7 7.0%
Total Records: 100

7 v. 10
Team 1 -> 73 73.0%
Team 2 -> 27 27.0%
Total Records: 100

8 v. 9
Team 1 -> 61 61.0%
Team 2 -> 39 39.0%
Total Records: 100

1 v. 16
Team 1 -> 1000 100.0%
Total Records: 1000

2 v. 15
Team 1 -> 999 99.9%
Team 2 -> 1 0.1%
Total Records: 1000

3 v. 14
Team 1 -> 995 99.5%
Team 2 -> 5 0.5%
Total Records: 1000

4 v. 13
Team 1 -> 991 99.1%
Team 2 -> 9 0.9%
Total Records: 1000

5 v. 12
Team 1 -> 961 96.1%
Team 2 -> 39 3.9%
Total Records: 1000

6 v. 11
Team 1 -> 896 89.6%
Team 2 -> 104 10.4%
Total Records: 1000

7 v. 10
Team 1 -> 773 77.3%
Team 2 -> 227 22.7%
Total Records: 1000

8 v. 9
Team 1 -> 578 57.8%
Team 2 -> 422 42.2%
Total Records: 1000

1 v. 16
Team 1 -> 10000 100.0%
Total Records: 10000

2 v. 15
Team 1 -> 9997 100.0%
Team 2 -> 3 0.0%
Total Records: 10000

3 v. 14
Team 1 -> 9974 99.7%
Team 2 -> 26 0.3%
Total Records: 10000

4 v. 13
Team 1 -> 9902 99.0%
Team 2 -> 98 1.0%
Total Records: 10000

5 v. 12
Team 1 -> 9600 96.0%
Team 2 -> 400 4.0%
Total Records: 10000

6 v. 11
Team 1 -> 8907 89.1%
Team 2 -> 1093 10.9%
Total Records: 10000

7 v. 10
Team 1 -> 7773 77.7%
Team 2 -> 2227 22.3%
Total Records: 10000

8 v. 9
Team 1 -> 6030 60.3%
Team 2 -> 3970 39.7%
Total Records: 10000

As you can see, not many underdogs winning. So I decided to come up with a historical formula based on these "facts" (via wikipedia):
  • The #1 seed has beaten the #16 seed all 100 times (100%).
  • The #2 seed has beaten the #15 seed 96 times (96%).
  • The #3 seed has beaten the #14 seed 85 times (85%).
  • The #4 seed has beaten the #13 seed 79 times (79%).
  • The #5 seed has beaten the #12 seed 66 times (66%).
  • The #6 seed has beaten the #11 seed 69 times (69%).
  • The #7 seed has beaten the #10 seed 61 times (61%).
  • The #8 seed has beaten the #9 seed 46 times (46%).

  • So, the #16 seed would have to overcome history and win 100% of the "draws". From there, the chances get a lot better for the underdog.

    This didn't fare out as well as I thought. Only the 7-10 and 8-9 matchups had the underdogs winning at least one game, and the 9 seed was the only underdog to win the matchup. Maybe once the brackets are released, I can pull things like PPG, PA, and different splits to make it team-dependent.




    No comments: