Final results - congratulations to Jordan, Jillian, and Debbie as our top three performers!

Benchmarks
1st
2nd
3rd
4th
5th
6th
Ave.
RPI
20-12
9-7
4-4
1-3
0-2
0-1
 .32
Kenpom.com
23-9
10-6
5-3
0-4
0-2
0-1
 .33
Seeding (RPI in Final Four)
23-9
10-6
5-3
0-4
0-2
0-1
 .33
 
Seeding-based (various in Final Four)
Dietz (friend's rec in Final Four)
23-9
10-6
5-3
0-4
0-2
0-1
 .33
Catalano (regular season matchups in FF)
23-9
10-6
5-3
0-4
0-2
0-1
 .33
 
Experts?
Johnson (self)
20-12
9-7
7-1
1-3
1-1
0-1
 .47
Childers (self)
23-9
10-6
5-3
1-3
1-1
0-1
 .45
Gerlach (ESPN and boyfriend)
19-13
8-8
3-5
2-2
1-1
0-1
 .41
Crim (self)
23-9
8-8
5-3
1-3
0-2
0-1
 .35
Lang (husband)
23-9
8-8
2-6
2-2
0-2
0-1
 .33
Thomas (friend)
25-7
10-6
3-5
0-4
0-2
0-1
 .29
Gilson (self)
23-9
9-7
4-4
0-4
0-2
0-1
 .29
LaPee (self)
22-10
9-7
4-4
0-4
0-2
0-1
 .29
Seneca (self)
23-9
11-5
2-6
0-4
0-2
0-1
 .27
Wheeler (self)
25-7
9-7
2-6
0-4
0-2
0-1
 .27
Ward (self)
23-9
8-8
2-6
0-4
0-2
0-1
 .25
 
Other
Racey (seeding 1st 2 rds, tourn. history after, error in 1st round)
24-8
10-6
4-4
1-3
1-1
0-1
 .44
Carns (teamrankings.com)
25-7
10-6
6-2
0-4
0-2
0-1
 .36
Bittner (web site's probabilities)
25-7
10-6
6-2
0-4
0-2
0-1
 .36
Davis (wins/losses)
20-12
9-7
4-4
2-2
1-1
0-1
 .36
Mason (recognition heuristic)
22-10
8-8
2-6
1-3
1-1
0-1
 .36
Snider (informal combo of various stats)
21-11
10-6
3-5
1-3
0-2
0-1
 .32
Lee (researched teams)
22-10
10-6
4-4
0-4
0-2
0-1
 .30
Price (various stats + husband)
23-9
10-6
3-5
0-4
0-2
0-1
 .29
Veverka (wins - losses, RPI if tie)
20-12
9-7
4-4
0-4
0-2
0-1
 .28
Whitaker (familiarity - schools and states)
21-11
9-7
3-5
0-4
0-2
0-1
 .27
Ikeda (coin flip)
13-19
4-12
2-6
1-3
0-2
0-1
 .19

The table shows the record of each person/strategy in each round of the tournament. The last column is one method of measuring performance and represents the average percent correct in each of the six rounds. Note that this method gives a lot of weight to picking the right teams in the later rounds.

The two benchmarks that I have found to be successful in the past are RPI and seeding. Kenpom.com is a new one that I am testing for the first time.

Note that this data is for a single year and thus only broad conclusions can be drawn. Small differences will not necessarily replicate. In 2002's tournament, RPI had a score of 0.38 and seeding had a score of 0.42 (RPI was used for the Final Four because these four teams all had the same #1 seeding). Both did better than all but one student in 2002. In 2004's tournament, RPI had a score of 0.41 and seeding a score of 0.36. These results were on par or better than nearly every other approach except, again, for one student.

2004 Results