Whenever a 12 seed is from a major conference and the 5 seed is from a mid-major, take the 12 seed.

The 12 seed is undefeated in this scenario going back to the last major conference re-alignment in 1991.

We are currently looking at other rules.  One that has our attention is when a major plays against another major and the one major plays in the finals of their conference tournament (Florida State last year, Syracuse the year they lost to Texas A&M in Gerry McNamara's senior year).

We'll let you know what we find.

Since the pod system was created in 2002, the advantage for the higher seeds has been staggering.  Now let's take this a step further.  Here is the make-up of a typical Elite 8.

Expect at least 3 one seeds.
Expect at least 6 teams to come from teams that have been to the Final 4 since 1985.
Expect one mid-major.
Expect one team that does none of the above (Missouri last year).

Expect 7 teams to have an RPI below 22 (last year, the highest RPI was 13).  The last team's RPI will be below 54 and more likely to be between 23-38.

So the question becomes, who are those outlier teams that may crash the party.  Here is the short list outside of teams that I have mentioned in the Final 4 posts:

Missouri
Maryland
BYU
West Virginia
Florida
Ga Tech
Butler

One thing that I've been reading about is how that since 1991, only one team has won the tournament without making it to the previous year's tournament (Syracuse, 2003).

That got me thinking more about my Final 4 probabilities.  Here is my collection:

Probability that:

Two teams are from one conference:  72%
One team from last year's Final 4:  72%
Probability that all the teams have an RPI <=22:  81%
Probability that all the teams are in the RPI Top 10:  19%
Probability that you will have 2 or more 1 seeds:  60%
Probability that you will have 3 or more 1 seeds:  16%

Here is what I just researched:

Probability that a team went to last year's tournament:  90%
Probability that one team in the Final 4 did not make last year's tournament:  36%

That last number is much higher than I would have anticipated.  However, knowing that numbers gives some hope to Georgetown and Kentucky, but this information would probably make you lean towards a:

Syracuse, Kansas are probably in.

Villanova and Michigan State are leading candidates for the 3rd slot.

Duke and Wisconsin lead for the 4th slot.  However, Duke is somewhat hurt by their high RPI.  The top 3 teams in the RPI have only made it once in the past 16 years, while 2 of the top 3 have made it 50% of the time.

Kentucky and Georgetown are still in consideration.

http://sports.espn.go.com/ncb/bracketology

Here is the bracket and our first round upsets from this bracket:

(9) Ga Tech over Illinois
(13) Utah State over Gonzaga* - game is in Spokane
(10) Florida State over Richmond
(11) Va Tech over Butler
(10) Dayton over N. Iowa
(11) Louisville over Texas A&M
(12) Marquette over Wake

Sweet 16
Kansas
Utah State
Baylor
Florida State

Syracuse
Texas
WVU
Purdue

Kentucky
Marquette
Wisconsin
Kansas State

Villanova
Vanderbilt
Ohio State
Duke

Elite 8
Kansas over Baylor
Syracuse over West Virginia
Wisconsin over Kentucky
Villanova over Ohio State- This one is 50-50, model is deadlocked.  Nova gets the edge to support our 1 team from previous final 4 rule

Championship
Kansas over Syracuse

Kansas
Duke
Villanova
Syracuse
Kentucky
Georgetown
Wisconsin

with Michigan State, Ohio State, Ga Tech, and Texas just outside our criteria.

Let's look at the 7 teams.

Kansas is clearly the strongest team at this point.  They seem to be a solid pick for the championship game.  Duke's loss to Georgetown still looms big.  Speaking of G-town, you have to like Monroe and Freeman, but you have to be skeptical with losses to South Florida and Rutgers.  I question whether or not they can win 2 in a row.

Villanova has a huge advantage in our models having gone to last year's final 4 and having a superstar like Scottie Reynolds.  At the same time, does Villanova beat American last year if that game isn't in Philadelphia?

Syracuse is probably the second strongest team behind Kansas.

Wisconsin is still playing well and I think they are a great dark horse pick for your bracket.

Kentucky - well they have Wall, Calipari, and rich boosters.  I will let you make your own conclusions.

There are several components that are involved in selecting the Elite 8 in my models.  Let me review some of the simpler rules of thumb that aren't in the model.

1) Pick at least 2 of the RPI Top 3 for sure.  These teams have at least a 63% chance of making the Elite 8 indivivdually, but only 5 out of 16 years have they all won in the same year.  14 of 16 years have 2 of these 3 made it.
2) Of the RPI Top 10, you typically get 4-7 of these teams of which we've just picked 3, so you need to pick 1 to 4 of these teams.  Historically, 5 is the mode.  We had 7 last year and the typical movement is 1, so we will go with 6.  We need to pick 3 of the remaining 7 teams.
3) Pick only 1 mid major and make sure they aren't in the RPI Top 10.
4) 80% of the time, expect a 6 seed or higher to make it.
5) The cumulative seeds will fall between 18 and 34 with 80% confidence.  We are expecting 22 to 28 right now.
6) It is equally as likely that there will be one team from last year's Elite 8 as there is 2 or more.  Only twice has there not been a team from last year's elite 8 going back to 1986.

1.  Kansas
2.  Kentucky
3.  Duke
4.  Syracuse
5.  Georgetown
6.  Wisconsin
7.  Michigan State
8.  Texas
9.  Villanova

Now, the bottom 3 are on the brink of following off the list.   Texas is currently 22nd in the RPI, our lower level threshold.  Michigan State and Villanova, are nearly ready to fall off the short list, based on their on-the court productivity.

If that happens, we are down to 6.  Duke doesn't pass the eye ball test with their lack of points in the pain.  They are reaching our upper limit on % of points from 3.

If you believe that 2 teams will come from one conference and that conference is the Big East, you are looking at Georgetown and Syracuse.  Kansas has our next strongest rating.  That leaves the last slot a battle between Wisconsin and Kentucky

There is a 92% chance of at least one 10 or higher seed in the Sweet 16.   There is a 64% chance of 2 or more.

There is a 96% chance of at least one 5 to 7 seed.  There is a 92% chance of at least 2.  There is a 64% chance of at least 3.  The mode is 5 (8 out of 25 years), but interestingly enough, only 8% of the time has there been 6 or more.

8 and 9 seeds making it to the Sweet 16 is very rare.  It has only happened 40% of the time and only 8% of the time do 2 make it.

In aggregate, 60% of the time you will 5-7 teams that are a 5 seed or higher in your Sweet 16.  16% of the time you have 2-4, and 24% of the time you will have 8 or 9.

The current 80% confidence band is 4-8.  When you look further at the change rate, last year we had 3 teams seeded 5th or lower make the Sweet 16.  84% of the time this number changes by 0 to 3 from the previous year.  So that opens a window of 0 to 6.

I am looking to have 5 or 6 teams seeded 5th or lower in my Sweet 16.  1 or 2 will be a 10 seed or higher and 3 or 4 will be a 5-7 seed.

The NCAA is showing their money greed again for the major conferences and they want to add 32 at-large teams to the tournament.  This is a really bad idea for several reasons.

1.  The regular season gets diluted even more.
2.  The excitement of the tournament will be lost if brackets either (a) have to be turned in sooner, or (b) has less turnaround time after this new first round.
3.  Most importantly, a 16 seed has never won.  This either shows that the tournament has it perfect, or they need to get rid of automatic bids.

If you were to take the RPI from 2/2, and simply took the top 64 teams, 24 would come from mid-majors.  If you used KenPom, you would have 21 teams.

Since the RPI originated in 1994, only three teams with an RPI higher than 96 have ever won an NCAA Tournament game.  Only 14 teams with an RPI higher than 64 have won a NCAA Tournament game, and it has only happened in 8 years out of 16.  It only has a 50-50 chance of happening.  That seems like it is perfectly calibrated in its current state.

I would love to hear other thoughts on this...