Sweet Sixteen Log5 Table

Using the latest Relative Ratings, I decided to run one million simulations of the tournament starting at the Sweet 16. Below is a Log5 table displaying the results. The win odds are similar to the Vegas odds, though my ratings result in Duke being the favorite, whereas Vegas has Kentucky as the favorite at 3 to 1 odds.

Team	S16	E8	F4	NC	Win
Northern Iowa	100.0%	48.4%	17.1%	5.2%	1.3%
Michigan St	100.0%	51.6%	19.0%	6.1%	1.6%
Tennessee	100.0%	32.8%	17.3%	5.8%	1.6%
Ohio St	100.0%	67.2%	46.7%	23.4%	10.0%
Syracuse	100.0%	73.5%	47.6%	32.1%	16.1%
Butler	100.0%	26.5%	10.7%	4.7%	1.3%
Xavier	100.0%	36.8%	12.4%	5.6%	1.7%
Kansas St	100.0%	63.2%	29.3%	17.0%	6.9%
Kentucky	100.0%	79.7%	49.7%	21.8%	12.2%
Cornell	100.0%	20.3%	6.2%	1.1%	0.3%
Washington	100.0%	32.9%	11.0%	2.7%	0.9%
West Virginia	100.0%	67.1%	33.1%	12.5%	6.1%
Duke	100.0%	78.8%	63.0%	46.1%	33.4%
Purdue	100.0%	21.2%	11.1%	4.8%	2.0%
Baylor	100.0%	63.2%	18.6%	8.5%	3.8%
St Mary's CA	100.0%	36.8%	7.4%	2.5%	0.8%

Bracketology: A First Attempt at Seedings (updated March 14th)

First some admissions: these seedings are based on a mixture of statistics, some of which the committee members use and some they probably do not. The result is that this bracket doesn't completely resemble others drifting around out there in the internet ether. However, one important statistic I use is predictive in nature, which means it is pretty good at guessing what will happen. So these seedings may not match what they would be if the season ended today, but are probably closer to what they will be when the conference tournament season is over. Still with me? Anyway, I have not applied bracket logic to these, so some of the lower seeds may be a spot off from what you may expect. Additionally, injuries have not been taken into account, which most notably affects Ohio St and Purdue. I have begun to make manual procedural bumps to more accurately reflect what I see as the likely seedings. Initially I wanted this to be a purely mathematical model, but then I decided that accounting for player injuries would be extremely difficult (I'd have to track injuries for every team who was in contention for an at large bid, discount the stats from that period a certain percentage based on that player's lost production, etc). I felt that a few manual seed changes were within the spirit of the exercise.

Updated through games of March 13th.

1. Kansas, Kentucky, Syracuse, Duke
2. West Virginia, Kansas St, Ohio St, Georgetown
3. Baylor, Villanova, Pittsburgh, Purdue
4. Wisconsin, New Mexico, Temple, Butler
5. Tennessee, Vanderbilt, Michigan St, BYU
6. Maryland, Texas A&M, Xavier, Richmond
7. Northern Iowa, California, Gonzaga, Marquette
8. Oklahoma St, Florida St, Notre Dame, Texas
9. UTEP, Clemson, UNLV, Louisville
10. St Mary's CA, Missouri, Utah St, Georgia Tech
11. Washington, Siena, San Diego St, Old Dominion
12. Cornell, Mississippi St, Wake Forest, Virginia Tech
13. Murray St, Wofford, New Mexico St, Sam Houston St
14. Oakland, Houston, Ohio, Montana
15. Vermont, Santa Barbara, North Texas, Morgan St
16. ETSU, Robert Morris, Lehigh, Winthrop, Ark Pine Bluff

Last Four In: San Diego St, Mississippi St, Wake Forest, Virginia Tech
Last Four Out: Minnesota, Arizona St, Memphis, Florida

Sunday Housekeeping

I have decided that instead of creating a new post each time I want to update the rankings, every morning I will update the March 4th post with the current relative rating rankings and the March 6th post with the current SAWP rankings.

Also look for a Bracketology post in the next few days, which will also be updated daily in the same manner as described above.

Sooner Rather Than Later

Since I posted the Top 50 rankings for the Relative Ratings a few days ago, I figured maybe I should post the Top 50 SAWP rankings. As a reminder, the SAWP rating only uses win/loss and location as inputs, which is similar to the RPI. SAWP will typically correlate much closer to the human polls than will the Relative Ratings.

I am not a big fan of win/loss rating systems, because although they are a reasonable zero-sum look at what happened, they aren't typically helpful in identifying what will happen. And in the end, isn't that what the NCAA tournament committee's job is? They have a product to sell, after all, and the best product should consist of the best 64 games. If RPI isn't a good indicator of what will happen, then why is it one of the main factors in determining which teams make the tournament? And now off the soapbox, I'll post the rankings.

Updated through games of March 12th

Top 50	Conf	W-L	SAWP	RPI/Rnk	SOS/Rnk
1. Kansas	B12	31-2	.7535	.665 / 1	.584 / 18
2. Kentucky	SEC	30-2	.7331	.644 / 4	.543 / 78
3. Syracuse	BE	28-4	.7256	.650 / 2	.585 / 17
4. West Virginia	BE	26-6	.7072	.644 / 3	.603 / 6
5. Duke	ACC	27-5	.7069	.640 / 6	.583 / 19
6. Purdue	B10	27-4	.7026	.621 / 9	.536 / 92
7. Kansas St	B12	25-6	.6991	.642 / 5	.610 / 3
8. New Mexico	MWC	29-4	.6990	.617 / 11	.526 / 114
9. Northern Iowa	MVC	28-4	.6898	.602 / 23	.501 / 164
10. Temple	A10	27-5	.6895	.611 / 15	.531 / 104
11. Butler	HL	28-4	.6879	.611 / 16	.527 / 111
12. Baylor	B12	24-7	.6811	.622 / 8	.582 / 21
13. Villanova	BE	24-7	.6807	.617 / 12	.564 / 43
14. Tennessee	SEC	25-7	.6784	.616 / 14	.568 / 40
15. BYU	MWC	28-5	.6774	.598 / 24	.504 / 157
16. Georgetown	BE	23-9	.6738	.632 / 7	.632 / 1
17. Pittsburgh	BE	24-8	.6737	.620 / 10	.589 / 13
18. Vanderbilt	SEC	23-7	.6721	.611 / 17	.566 / 41
19. UTEP	CUSA	26-5	.6703	.594 / 29	.510 / 140
20. Xavier	A10	24-7	.6694	.604 / 21	.547 / 70
21. Ohio St	B10	25-7	.6656	.594 / 28	.530 / 107
22. Utah St	WAC	26-6	.6644	.590 / 32	.520 / 121
23. St Mary's CA	WCC	25-5	.6615	.590 / 33	.506 / 152
24. Richmond	A10	25-7	.6603	.593 / 31	.531 / 101
25. Maryland	ACC	22-8	.6602	.606 / 18	.575 / 30
26. Gonzaga	WCC	25-6	.6596	.585 / 40	.506 / 150
27. Cornell	Ivy	25-4	.6590	.568 / 59	.459 / 271
28. Texas A&M	B12	22-9	.6563	.616 / 13	.606 / 5
29. Wisconsin	B10	23-8	.6546	.606 / 19	.575 / 31
30. UNLV	MWC	24-7	.6543	.586 / 38	.516 / 128
31. Siena	MAAC	27-6	.6537	.580 / 46	.496 / 177
32. Murray St	OVC	28-4	.6524	.553 / 76	.425 / 318
33. Michigan St	B10	24-8	.6522	.596 / 25	.549 / 65
34. Texas	B12	24-9	.6490	.595 / 26	.549 / 66
35. San Diego St	MWC	22-8	.6460	.586 / 39	.542 / 79
36. Old Dominion	CAA	26-8	.6455	.585 / 41	.531 / 100
37. California	P10	23-9	.6448	.605 / 20	.589 / 14
38. Oklahoma St	B12	22-10	.6392	.595 / 27	.576 / 28
39. Rhode Island	A10	23-8	.6376	.584 / 43	.533 / 98
40. Notre Dame	BE	23-11	.6367	.587 / 37	.561 / 49
41. Florida St	ACC	22-9	.6356	.588 / 35	.553 / 62
42. Virginia Tech	ACC	23-8	.6351	.567 / 60	.494 / 184
43. Louisville	BE	20-11	.6346	.602 / 22	.610 / 4
44. Marquette	BE	22-11	.6344	.583 / 44	.558 / 53
45. Missouri	B12	22-10	.6341	.584 / 42	.552 / 63
46. Washington	P10	23-9	.6325	.580 / 45	.532 / 99
47. Clemson	ACC	21-10	.6323	.593 / 30	.579 / 24
48. UAB	CUSA	23-8	.6304	.579 / 48	.530 / 106
49. Wichita St	MVC	24-9	.6285	.572 / 55	.519 / 123
50. Wofford	SC	25-8	.6265	.545 / 86	.459 / 269

Notes: The RPI rating shown may not precisely equal other sources. It is close enough that I am comfortable in providing it simply as a comparator to SAWP.

A Return to March Madness

As my readership is extremely small to non-existent, I have failed to keep consistent with my posting. However, with March here and college basketball entering the conference tournament season, I wanted to post my Top 25 Top 50 rankings which are based on my week adjusted relative ratings.

Updated through games of March 21st

Top 50	Conf	W-L	Pyth	AdjO/Rnk	AdjD/Rnk
1. Duke	ACC	31-5	.9813	80.4 / 3	58.6 / 1
2. Kansas	B12	33-3	.9808	82.2 / 1	60 / 4
3. Syracuse	BE	30-4	.9675	81 / 2	61.7 / 16
4. Kentucky	SEC	34-2	.9612	79.6 / 5	61.6 / 14
5. Ohio St	B10	29-7	.9553	76.7 / 17	60.1 / 5
6. BYU	MWC	29-6	.9526	79.6 / 6	62.6 / 26
7. Kansas St	B12	27-7	.9513	79.8 / 4	62.9 / 32
8. West Virginia	BE	29-6	.9488	77 / 15	60.9 / 10
9. Wisconsin	B10	24-9	.9481	74.2 / 45	58.8 / 2
10. Baylor	B12	26-7	.9399	78 / 10	62.6 / 25
11. Maryland	ACC	23-9	.9384	78.7 / 8	63.3 / 40
12. Texas	B12	24-10	.9355	79.1 / 7	63.8 / 50
13. Missouri	B12	23-11	.9340	77.4 / 11	62.6 / 27
14. Purdue	B10	29-5	.9338	75.1 / 32	60.8 / 8
15. Georgetown	BE	23-11	.9277	76.5 / 18	62.4 / 23
16. Utah St	WAC	26-8	.9271	75.1 / 30	61.3 / 13
17. Clemson	ACC	21-11	.9247	75.9 / 24	62.1 / 21
18. California	P10	24-11	.9223	77.1 / 14	63.2 / 38
19. Xavier	A10	26-8	.9193	77.3 / 13	63.6 / 47
20. Butler	HL	30-4	.9152	73.9 / 56	61.1 / 12
21. Florida St	ACC	22-10	.9134	73.6 / 58	60.9 / 9
22. Texas A&M	B12	23-10	.9132	74.9 / 38	62 / 20
23. Tennessee	SEC	27-8	.9128	75.7 / 25	62.7 / 29
24. Villanova	BE	25-8	.9096	78.6 / 9	65.3 / 84
25. Minnesota	B10	21-14	.9091	75 / 37	62.3 / 22
26. Michigan St	B10	26-8	.9067	75.1 / 31	62.6 / 28
27. Old Dominion	CAA	27-9	.9028	72.3 / 74	60.5 / 7
28. Pittsburgh	BE	25-9	.9015	73.7 / 57	61.8 / 17
29. St Mary's CA	WCC	27-5	.9012	76.3 / 20	63.9 / 53
30. Washington	P10	26-9	.9010	77.4 / 12	64.8 / 71
31. Northern Iowa	MVC	30-4	.9007	71 / 99	59.5 / 3
32. Temple	A10	29-6	.9000	71.7 / 83	60.2 / 6
33. UTEP	CUSA	26-7	.8988	75.5 / 26	63.4 / 42
34. Marquette	BE	22-12	.8982	75.1 / 33	63.1 / 34
35. Vanderbilt	SEC	23-9	.8960	76.8 / 16	64.6 / 65
36. Georgia Tech	ACC	22-13	.8955	75 / 35	63.1 / 36
37. Murray St	OVC	29-5	.8896	74.3 / 44	62.9 / 31
38. Memphis	CUSA	24-10	.8848	75.2 / 29	63.9 / 52
39. Virginia Tech	ACC	24-8	.8832	74.7 / 40	63.5 / 44
40. Arizona St	P10	22-11	.8829	71.6 / 85	61 / 11
41. UNLV	MWC	24-9	.8793	74 / 51	63.1 / 37
42. Dayton	A10	21-12	.8785	73.3 / 61	62.6 / 24
43. Louisville	BE	20-13	.8767	76.1 / 22	65 / 74
44. Mississippi	SEC	23-10	.8765	76.4 / 19	65.3 / 83
45. Mississippi St	SEC	24-12	.8760	74.2 / 47	63.5 / 43
46. New Mexico	MWC	30-5	.8717	75 / 34	64.3 / 62
47. Oklahoma St	B12	22-11	.8628	74.8 / 39	64.6 / 64
48. Cornell	Ivy	27-4	.8628	74 / 52	63.9 / 51
49. Florida	SEC	21-13	.8618	74.2 / 46	64.1 / 58
50. San Diego St	MWC	23-9	.8603	72.6 / 70	62.8 / 30

Note: Teams highlighted in red remain active in the NCAA tournament