The S-Factor: Tuesday, February 15

Now in: Old Dominion
Now out: Southern California

Conferences with multiple bids:
Big East: 10
Big Ten: 7
ACC: 6
Big 12: 5
SEC: 5
Atlantic 10: 4
PAC-10: 2
Colonial: 2

About the S-Factor

The College Women's Hoops S-Factor is an algorithm that attempts to rank women's basketball teams according to their likelihood of being selected for participation in the NCAA Women's Basketball Championship Tournament, and it attempts to predict the seed given to each team.

It is a bracketology algorithm.

The S-Factor ranks teams based exclusively on wins and losses, and the various ways to gauge the quality of each win and loss. Specifically, the S-Factor algorithm uses the following information, listed in order of magnitude of weighting.

- Conference W-L % weighted against conference strength
- RPI

- Record against top 25 RPI teams
- Elite conference factor (more here)
- Record against top 50 RPI teams
- Overall record against NCAA Div. I teams
- Conference tournament performance
- Not having losses to sub-100 RPI teams

- Not having losses to sub-200 RPI teams

The S-Factor algorithm does not take into account any of the following:

- Number of points
- Margin of victory
- Whether or not games go to overtime
- Rebounds, assists, blocks, turnovers, etc.
- Shooting percentages
- Free-throw percentages
- Key players who may be injured
- Any individual statistic

The goal of the S-Factor is not to predict game outcomes or establish betting lines. It's not even very good at predicting tournament winners. I am only interested in trying to figure out which teams will get selected to the tournament by the NCAA Women's Selection Committee.

The College Women's Hoops S-Factor correctly predicted 61 out of the 64 teams in 2011, and it predicted the seeding of 80% of the teams within one seed of the actual result (51 out of 64). It was the best numerical method in terms of tournament selection prediction and seeding predictions (for more, see this post).

In 2012, my computer blew up and I lost my data.

In 2013, (after rebuilding the database) the S-Factor correctly predicted 62 out of the 64 teams, including 25 teams seeded exactly right and 44 teams seeded within one of the actual result. Once again it was the best bracket prediction performed by a numerical method (2013 Bracketology-ology).

In 2014, the S-Factor correctly predicted 61 out of the 64 teams, including 26 teams seeded exactly right and 47 teams seeded within one of the actual result. (2014 Metabracketology)

The S-Factor

Pages

Tuesday, February 15, 2011

Tuesday, February 15

No comments:

Post a Comment