January 5 Update: Polls and Outliers

The table below shows the S-Factor in relation to the AP poll, the RPI rankings, and last year's S-Factor formulation.

Team	AP	RPI	2015 S-Factor	2014 S-Factor
South Carolina	1	12	10	8
Connecticut	2	2	7	4
Texas	3	8	4	5
Notre Dame	4	1	2	1
Baylor	5	13	17	14
Louisville	6	8	1	2
Tennessee	7	5	24	16
North Carolina	8	6	12	12
Texas A&M	9	36	19	15
Kentucky	10	3	5	3
Oregon State	11	29	13	18
Maryland	12	16	3	6
Duke	13	17	14	23
Mississippi State	14	35	11	10
Stanford	15	44	26	36
Oklahoma State	16	84	70	59
Iowa	17	10	8	11
Arizona State	18	4	6	7
Nebraska	19	26	22	31
Georgia	20	45	32	29
Syracuse	21	67	48	65
Princeton	22	19	44	32
Minnesota	23	15	9	9
Rutgers	24	53	63	46
Western Kentucky	25	23	51	43
Washington	26	18	21	25
St. John's	27	30	29	17
Florida State	29	28	18	24
Washington State	30	49	16	20
California	32	25	25	28
Ohio State	33	31	15	21
Northwestern	34	37	20	19
Green Bay	35	7	33	27
South Florida	38	20	31	22
George Washington	40	22	23	13

The S-Factor is in rough agreement with the voters of the AP regarding many of the teams, but there are some interesting outliers. The S-Factor ranks Oklahoma State, Syracuse, and Rutgers far below their positions in the AP poll, and this is mostly due to their poor RPI so far. Syracuse and Rutgers have no wins better than 63 RPI Arkansas so far, while Oklahoma State has feasted upon cupcake teams en route to the 275th most difficult schedule in the 349-team NCAA Division 1.

The S-Factor thinks quite highly of Maryland, Louisville and Minnesota. Maryland and Louisville rank higher than the AP poll mainly because of their three top-25 wins each. No other team has more than three top-25 wins, though Connecticut, Kentucky and Vanderbilt also have 3. Minnesota ranks highly because they've played and won more conference games than any other team in an elite conference. While it is most definitely a good thing to win conference games, this early Big Ten schedule just means that Minnesota's numeric advantage in the S-Factor will ebb by the time the season is over and they will fall more or less in line with the AP ranking.

Now in:
South Florida
Iowa State
Oklahoma
Wichita State
Western Michigan
Albany
Liberty
Southern U.
North Dakota St.

Now out:
Arkansas
Seton Hall
West Virginia
Indiana State
Akron
South Dakota St.
Maine
High Point
Prairie View A&M

Conferences with multiple bids:
Big Ten: 8
SEC: 7
PAC 12: 7
ACC: 6
Big 12: 5
Atlantic 10: 2
American: 2
Atlantic Sun: 2
West Coast: 2

Why the S-Factor is Cramming Ten Teams In From One Conference

The College Women's Hoops S-Factor shows ten of the sixteen Big East teams getting into the dance: Connecticut, DePaul, Georgetown, West Virginia, Marquette, Notre Dame, Syracuse, Louisville, St. John's and Rutgers. This, frankly, is absurd.

The Big East chart shows the rigid stratification of the conference. Basically the Big East can be divided into 4 groups: the Connecticut Group, the Ranked Group, the Middling Group and the Big Least.

The Big Least is composed of six teams (Villanova, Providence, Seton Hall, Pittsburgh, Cincinnati and South Florida) who have gone a combined 6-43 in conference games. They have achieved zero wins against teams from the other groups.

The Connecticut Group is composed of three teams (Connecticut, DePaul, Notre Dame) who have combined for only one loss in an intragroup game (Notre Dame losing to Connecticut).

The most mixing occurs in the two middle groups; the Ranked Group (Georgetown, Marquette, West Virginia), and the Middling Group (Rutgers, St. John's, Syracuse, Louisville). The Middling Group has achieved two wins over the Ranked Group (Rutgers over Georgetown, Syracuse over Marquette). These two wins have kept the Middling Group as a whole above the at-large bubble line in the College Women's Hoops S-Factor.

Here is a table of the relative records against each other:

	C.Gr.	R.Gr.	M.Gr.	Least
Connecticut Group	1-1	4-0	7-0	13-0
Ranked Group	0-4	3-3	4-2	11-0
Middling Group	0-7	2-4	4-4	13-0
The Big Least	0-13	0-11	0-13	6-6

Besides those two wins previously discussed, there is no intergroup mixing of wins. And it is this oddly stratified configuration that is tricking the S-Factor into thinking there will be ten Big East teams in the tournament.


Now in: BYU, St. John's
Now out: Wyoming, Boston College

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

The Outliers

It's hard to tell which teams selected by the College Women's Hoops S-Factor are mistakes and which ones are legitimate since there is only one confirmation test per year. But when one compares the S-Factor to other rankings generated by humans and computers, one gets a sense of where the outliers are.

Clearly the S-Factor is having problems seeing the merit of teams from mid-major conferences. The Horizon League's Wisconsin-Green Bay is ranked in both national polls, Charlie Creme's week old bracket has them in as a 5 seed, and a collection of models (Massey Ratings, Sagarin Ratings, Steven Jens and Sonny Moore) puts the Phoenix somewhere around a 7 seed. Yet the S-Factor huffily bestows an automatic-only seed of 11. Bowling Green, out of the Mid America Conference, winners of 14 games in a row and very close to being nationally ranked, is getting an average of a 7 seed by the computers (Steven Jens thinks they're the 16th best team in the country). But the S-Factor deigns only to give them a 12 seed, and nothing at all if they lose their conference tournament.

There are also teams that the S-Factor is infatuated with. Michigan, despite dropping their last two games, shows up high in the S-Factor with a predicted 6 seed, which is 4 seeds higher than the nearest outside computer model (Sagarin) would have them. (The S-Factor really loves those wins over Ohio State and Iowa). Creme's bracket doesn't even show Michigan in the "last four out" category. Another Big Ten team much loved by the S-Factor is Penn State, a team both humans and computers agree should be seeded somewhere south of an 8 seed (if at all). But the S-Factor puts them as the highest of the 5 seeds.

Some teams that seem too highly ranked with regard to human polls and brackets are actually mostly in line with other computer rankings. DePaul (2), Marquette (3), Texas Tech (5) and Louisville (7) are all teams that get more respect from number crunching than from human prognosticators, and the S-Factor sides more with the computers. The S-Factor also agrees with computers when humans appear to be overly sympathetic to some teams, like Ohio State (9), Kentucky (9), Syracuse (10) and Boston College (10).

And then there's the curious case of Oklahoma. Charlie Creme has them as a 6 seed. The polls would have them in with a 4 seed. Looks good for them, right? But according to RPI, the Sooners aren't anywhere near being in the tournament. The computer models Massey, Sagarin, and Jens all put them about where the S-Factor does around the mid-forties. But Sonny Moore ranks them 12th in the country, which would correspond to a 3 seed. Crazy!


Now in: (same as yesterday)
Now out: (same as yesterday)

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