Re: South seedings attempt

On Sun, 16 Apr 2006, Adarichev wrote:

I still don't understand the argument for why Texas St. should be
seeded ahead of Wash U. The last 2 posters argue for this and the
score reporter's South Regional page also has Texas at #2 in
preliminary seedings, but it simply doesn't make sense. Sure, State's
RRI is higher, but this measure isn't valid because they only went to 2
tourneys all year and have only played 6 games all season against teams
that have qualified for Regionals. They have good wins against North
Texas and LSU, but their RRI is ballooned by blowout wins against teams
like Texas Dallas, Texas C and Texas Christian.

Argument for state ahead of wash U:

(1) State beat LSU
(2) LSU beat Wash U

Argument for wash U ahead of state

(1) wash U beat kansas, kansas beat state

Counterpoint

(1) state beat UNT, who beat kansas 13-6 in most recent matchup, negating the argument for wash U ahead of state.


They are a good team, no doubt, but if they are seeded 2nd, they are
simply being rewarded for playing a short schedule. That said, no
matter where Texas State is seeded, North Texas should be seeded right
behind them.

It seems that you wants to punish state for playing a short schedule/reward washington for playing a long schedule. It simply doesnt make sense to me to say that washington must be ahead of texas state. And as we're about to find out, it may make sense to seed UNT not immediately behind texas state.


Also, LSU has to be seeded below Kansas. After a quick glace at score
reporter, they have 7 common opponents and Kansas performed better 6/7
times.RRI echoes this fact. I think this has to be the order at the
top:

As pointed out above, LSU has to be seeded above washington because of head to head... one of the two (lsu over washington, or kansas over lsu) is going to have to be broken. My initial instinct is to break the common opponents rather than head to head. but that's just my bias.


1. Texas
2. Wash U
3. Texas State
4. UNT
5. Kansas
6. LSU
...

-Adarichev, Wash U #31


Lets look at this seeding and count the number of head to head matchups that are "ignored". Counting violated head to head matchups is arbitrary, but it appears to be a concrete way to evaluate seedings and puts the emphasis on games that were played between relevant teams rather than the length of their season.

1) Texas 0
2) Washington 2 -- lost to lsu, UNT
3) texas state 1 -- lost to kansas
4) UNT 1 -- lost to arkansas
5) Kansas 0
6) LSU 0
7) arkansas 0
....

So 4 head to head matchups are "broken", not a great indication for the top of the seedings.

Now lets look at Zong's seeding:

1) texas 0
2) texas state 1 -- lost to kansas
3) unt 1 -- lost to arkansas
4) lsu 0
5) washington 0
6) kansas 0
7) arkansas 0

Breaking only 2 head to head matchups

Considering Malone's seeding we have:

1) texas 0
2) state/washington 1/2 -- lost to kansas/lsu, unt
3) state/washington 1/2 -- lost to kansas/lsu, unt
4) unt/kansas 1/0 -- lost to arkansas/nothing
5) unt/kansas 1/0 -- lost to arkansas/nothing
6) lsu
7) arkansas

Breaking 5 head to head matchups.

One alternative which may be suggested next is

1) texas 0
2) state/washington 1/2 -- lost to kansas/lsu, unt
3) state/washington 1/2 -- lost to kansas/lsu, unt
4) kansas 0
5) arkansas 0
6) unt 0
7) lsu 0


breaking 4 head to head matchups.

Another alternative

1) texas 0
2) washington 2 -- lost to lsu, UNT
3) kansas 0
4) state/arkansas 0
5) state/arkansas/lsu 0
6) arkansas/lsu/unt 0
7) lsu/unt 0


Where 2 head to head results are violated. (any of the teams in 4-7 can be chosen appropriately without changing that number).


So based off looking only at head to head matchups, the seeding of the top 7 should be either

1 texas
2 texas state
3 unt
4 lsu
5 washington
6 kansas
7 arkansas

or

1 texas
2 washington
3 kansas
4 state/arkansas
5 state/arkansas/lsu
6 arkansas/lsu/unt
7 lsu/unt

The bottom 8 are much harder to do using this metric since there aren't enough head to head matchups between sections to consider (a&m over oklahoma is the only one i see).

flame away.

allen
.

Relevant Pages