The following was proposed by Michael Schulz of LSU on the bama
section blog (http://bamasecs.wordpress.com/) and looks to be pretty
good
1. Texas (H-H Win over Kansas)
2. Kansas (Best Regular Season/RRI)
3. UNT (2nd in TX Section)
4. Texas State (3rd in TX + H-H over Arkansas)
5. Arkansas (2nd in OZ + H-H over A&M)
6. Texas A&M (4th in TX + H-H over LSU)
7. LSU (1st in BAMA + H-H W&L to Arkansas)
8. Wash U (3rd in OZ)
9. Truman (4th in OZ)
10. Missourri (5th in OZ)
11. Ole Miss - These guys could be 9th but Truman’s RRI is way better
but got 4th in the section so Truman has to be behind Wash U.
Missourri and Ole Miss could go either way but to avoid rematches.
12. Rice - (5th TX)
13. Harding - (6th OZ) - Rice and Harding could switch.
14. Miss St. - (3rd BAMA) Harding has a better record against common
opponents and better RRI.
15. Houston - (6th TX)
16. UTA - (7th TX)
This gives us:
Texas
UTA
Wash U
Truman
Arkansas
Rice
Tx St
Harding
UNT
Miss St
Tx AM
Ole Miss
LSU
Missouri
Kansas
Houston
This leaves only 2 inter-sectional matchups Tx-UTa and Wash U-Truman
in the first round which is pretty good considering 13 teams come from
2 sections.
You started it.
Before you get seeds from me, you get a statement about what seedings
are good for:
You can either view seedings as predictive of the results of the
upcoming tournament or reflective of the results of the previous
played games. Given the UPA requirement that sectional (regional)
finish must be maintained, seeding at regionals (nationals) must be
reflective rather than predictive.
Before you get seeds, you get to read how I got the seedings -- for
me, the process is more important than the outcome (gee, have I said
that before?). The following explanation captures the key steps and
tries to explain the rationale for subtle parts of the process.
(1) consider the highest remaining team from each section
(2) compare these teams to eachother in a mini tournament (similar to
what is done on the UPA SRT tool in the team grid)
the points for consideration are, in order:
(a) head to head record
(b) record against common opponents at regionals that have not
already been seeded
(c) record against common opponents not at regionals
(d) RRI, last years results, flip a coin -- at this point
there really isnt much useful information. Get teams to play more
games!
(3) after playing out this mini tournament, see if any team has a
unique best record. If that team exists, they get the next seed. If
that team does not exist, then there are two possiblities
(a) if every matchup was decided with the same criteria (i.e.
head to head) then consider the next criteria (common opponents) to
make the decision
(b) if matchups were decided using different criteria, then
nullify all of the lowest criteria used to make decisions (i.e. 1
matchup with head to head, 2 with rri, then the rri matchups become
draws)
(4) repeat this process until all teams are seeded.
There are a couple of important clarifications when applying this
algorithm
(1) When comparing common opponents, compare the winning percentage
against each individual team (i.e. aTm goes 2-1 against Davidson while
LSU goes 1-0 against Davidson, then LSU would 'win' that matchup based
on that common opponent)
(2) Placing higher than another team at sectionals negates all
previous victories/losses (i.e. if texas went 0-3 against aTm this
year, then because texas placed higher than atm at sectionals, that
record would be interpreted as 1-0).
(3) do not consider common opponents that have already been seeded.
Basically, using Arkansas's victory over Texas as an argument to seed
Arkansas ahead of aTm is also an argument to seed arkansas above
Texas. If texas has already been seeded, then we the argument is no
longer compelling (and we have already decided that Arkansas is below
texas, so the whole transitivity relationship begins to fall apart).
(4) Consider common opponents at regionals separately from common
opponents not at regionals for 2 reasons. First, it gives us another
layer of 'tie breaks'. Second, a common opponent at the tournament
will be directly impacted by the seedings so results with that
opponent should weigh more heavily than results against opponents not
at the tournament.
And now you get the seedings, with the short explanation for why each
seed was selected.
(1) Texas
o Texas over Kansas head to head
o Texas over LSU based on common opponents 2-1 (GA Tech, atm in
favor of texas, Arkansas in favor of LSU)
- note this is tied 1-1 for teams at regionals, and 1-0 for
teams not at regionals
(2) Kansas
o Kansas over UNT based on common opponents UCSD, arkansas
- again, arkansas at regionals is enough to stop consideration.
UCSD is just gravy
o Kansas over LSU based on common opponents OK State, arkansas,
notredame, michigan state
- see the comment above
(3) Arkansas
o arkansas over UNT based on head to head
o arkansas over LSU based on common opponents Missouri state and
atm
- atm is at regionals, so missouri state is just extra
(4) UNT
o LSU over UNT based on common opponent Davidson
- not at regionals
o Washington over LSU based on common opponents 3-1 (Notre
Dame, Michigan St, Missouri state in favor of wash, davidson in favor
of LSU).
- no teams at regionals in the common opponent list
o UNT over washington based on common opponent Tx State.
- tx state at regionals, carries more weight than other options
This is the first tricky one. LSU and UNT have a common opponent
arkansas, but Arkansas has already been seeded so should be ignored.
LSU over UNT is based on team not at regionals, Washington over LSU is
based on teams not at regionals. UNT over washington is based on a
team at regionals. So the UNT edge remains at the 'top' and we break
one of the latter two. End result, UNT as the 4 seed.
(5) tx state
o Washington over LSU based on common opponents (3-1 as above)
o Tx State over washington based on head to head
o tx state over lsu based on common opponents atm at regionals,
common opponents Ohio St, OK St not at regionals
(6) washington
o washington over lsu as above
o washington over atm on common opponents 2-1 (OK, Missouri in
favor of washington, davidson for atm)
- again, Missouri at regionals is enough at 1-0. the other two
are a wash
o atm over lsu head to head
(7) truman state
o truman over lsu based on common opponents emory, missour state,
ok state
- none at regionals
o truman over atm based on common opponent OK state, missouri
- missouri at regionals stops discussion
o atm over lsu head to head
(8) atm
o atm over lsu head to head
o missouri over atm head to head
o lsu over missouri head to head
====
o atm over lsu on rri (common opponents are identical records)
o atm over missouri on common opponent grinnel
- not at regionals
o missouri over lsu on common opponents ok state, missouri state
- not at regionals
This one is also tricky given the chain of head to heads. at the end
of the day, atm has claims over the other two teams on weaker
conditions so they get the seed.
(9) lsu
o lsu over missouri head to head
o lsu over rice common opponents 2-1 (new mexico, davidson for lsu,
ok state for rice)
- none at regionals
o missouri over rice on RRI
(10) missouri
o missouri over rice on rri
o missouri over mississippi on rri
o mississippi over rice on rri
(11) mississippi
o mississippi over rice on rri
o rice over harding on rri
o mississipppi over harding head to head
(12) rice
o rice over harding on rri
o harding over missippi state on rri
o rice over miss. state on rri
(13) ms state
o ms state over harding on rri
o harding over houston on rri
(14) harding
o harding over houston on rri
(15) houston
(16) UT-Arlington
For the TL;DR folks:
1) Texas
2) Kansas
3) Arkansas
4) UNT
5) tx state
6) washington
7) truman
8) atm
9) lsu
10) missouri
11) mississippi
12) rice
13) ms state
14) harding
15) houston
16) ut-arlington
These seeds give an unfortunate number of sectionals rematches and put
the texas and ozarks sections on opposite sides of the bracket
(assuming a 16 team bracket). In order to resolve those issues, I
advocate swapping the 3/4 seeds, the 7/8 seeds, and the 12/13 seeds.
This leaves only one inter-sectional matchup between two teams from
the section that has 7 teams at regionals (texas - UT Arlington).
ps. Mike, is this a good start on reaching 100 by Tuesday?
http://www.upa.org/scores/scores.cgi?div=127&page=3&tourn=6418
-ajax
1. Auburn
2-16. everyone else.
Damnit Ajax. You caught me!
-DW
Bravo C-bag. Now go finish your degree.
Totally agreed. Next year I say we petition to remove the 'Bama
section from Regionals. They are just hogging up 3 whole spots. Now
what about that 17th seed.
- Zeke
So you make this long post with a thorough (though still arbitrary and
based on which factors you deem to be more important than others)
analysis of seedings, and then you conclude it by saying basically -
this doesn't work so switch these 6 teams around. If you just move
teams around to fit into the part of the framework not accounted for
in your system, then what is the point of the system? Despite the
hard work put into the system you provide, since it is incomplete, are
you not in the end just placing teams as you see fit? Your omission
of one part of the whole (avoiding sectionals rematches) necessarily
changes all of the results of your system. For instance, if you were
to omit some other part of the system (sectional wins trump all other
meetings or ignoring already seeded teams) the results would be very
different.
ps. I am doing my part for "100 or Bust 2009"
Allen,
this is your most important tool to make your algorithm manageable.
But this is the place where your system is flawed. If an upset happens
at sectionals you punish the upsettee severely (give them a seed very
far below the seed they would have gotten without the upset) and
reward the upsetter only barely (give them a seed just slightly above
the seed they would have gotten without the upset).
Punishing the upsettee harshly may seem ok, but this also punishes the
first round opponent of that team by having to play a much better team
than earned. Yes, our regional formats are fairly robust against
unbalanced (in the sense of predictive seeding), but if you argue that
way you may as well seed alphabetically or randomly.
The problem at hand is that the UPA seeding tries to do two things:
1. Seed teams according to their accomplishments all year
2. Keep the order decided at sectionals to mimick a big tournament
consisting of sectionals/regionals/nationals
These two things don't work together if severe upsets happen. You
could get 1 if you dropped the requirement of seeding by placement of
previous stages and used the results from previous stages merely as
most recent (and thus important) full strength comparison between two
teams.
Or you could completely go with the second philosophy, where teams in
the region are completely seeded before sectionals, and if upsets
happen the two teams merely switch seeds in the next stage.
The UPA currently wants a mixture of both... and I like Tarr's
philosophy on this: Seed all teams according to philosophy 1 ("true
seeds"). Then, if this violates the sectionals-regionals rule due to
an upset at sectionals, seed both these teams as close as possible to
the average of their combined strength (look at their combined record,
normalized for number of games played per team and treat them as one
team) in the order inherited from sectionals.
Flo.
My take on the seeding instructions from the UPA is that they are
underspecified and trying to do two contradictory things at once,
which you have identified. Given the strict "Obey sectional finish"
requirement, it seems to me that the UPA is at least, implicitly,
stating that seeding teams according to the accomplishments of the
year is the correct thing to do and that "predictive seeding" is not
the right answer. Consequently, I do think that a mechanical
mechanism for seeding that can be applied by any diligent person is
appropriate. Which mechanism should we use? I find that to be an
interesting, and open, question.
I think it is important and reasonable to consider the next team from
each section in isolation rather than in the context of the lower
finishing teams because of the very real possiblity (Delaware* from
the Colonial section this year) of a good team being "upset" and not
qualifying for regionals. Does the fact that Delaware was expected to
be the 3 seed at ME regionals justify pushing the teams that upset
Delaware up into the high seeds of regionals even though Delaware is
not at that tournament? To me, the answer is no. I don't think the
seed for the top seed from the colonial section should not change
dramatically based on how many teams from the section qualify for
regionals.
>
> Punishing the upsettee harshly may seem ok, but this also punishes the
> first round opponent of that team by having to play a much better team
> than earned. Yes, our regional formats are fairly robust against
> unbalanced (in the sense of predictive seeding), but if you argue that
> way you may as well seed alphabetically or randomly.
Honestly, I would not object to alphabetic or random seeding,
especially for the pool play to bracket formats.
>
> The problem at hand is that the UPA seeding tries to do two things:
> 1. Seed teams according to their accomplishments all year
> 2. Keep the order decided at sectionals to mimick a big tournament
> consisting of sectionals/regionals/nationals
>
> These two things don't work together if severe upsets happen. You
> could get 1 if you dropped the requirement of seeding by placement of
> previous stages and used the results from previous stages merely as
> most recent (and thus important) full strength comparison between two
> teams.
>
I disagree with the assessment that the most recent game is the most
important comparison between two teams. How do you compare a game at
stanford with perfect weather to a game at fool's fest with 35 mph
sustained winds in which every game is decided by the flip?
(hypothetical). Fool's fest is more recent, but the reseults are
clearly skewed by weather. If you discard the 35mph winds, what about
30? 20? at what point are the wins 'acceptable' to consider?
> Or you could completely go with the second philosophy, where teams in
> the region are completely seeded before sectionals, and if upsets
> happen the two teams merely switch seeds in the next stage.
>
> The UPA currently wants a mixture of both... and I like Tarr's
> philosophy on this: Seed all teams according to philosophy 1 ("true
> seeds"). Then, if this violates the sectionals-regionals rule due to
> an upset at sectionals, seed both these teams as close as possible to
> the average of their combined strength (look at their combined record,
> normalized for number of games played per team and treat them as one
> team) in the order inherited from sectionals.
How do you seed teams according to philosophy 1 ("true seeds"
reflective of their results from this year)? My candidate mechanism
for doing that seeding would be to go back to my initial description
and replace "next team from each section" with "all candidate teams
for the next seed". When seeding regionals, I currently define the
candidate team from each section to be the highest available finisher
from that section; other definitions are viable. Do you have another
candidate that is better?
allen
* I am basing all discussion of Delaware on the fact that they were
mentioned in the C1 team list at the end of the year. If their actual
play did not match that preseason evaluation, then please treat them
as a strawman example.
> So you make this long post with a thorough (though still arbitrary and
> based on which factors you deem to be more important than others)
> analysis of seedings, and then you conclude it by saying basically -
> this doesn't work so switch these 6 teams around. If you just move
> teams around to fit into the part of the framework not accounted for
> in your system, then what is the point of the system? Despite the
> hard ...
>
(1) I'm more interested in the seeding mechanism than the results of
the seeding. I freely acknowledge that the factors I choose to
consider are somewhat arbitrary. After all, there are many factors
(total games played, points scored, point differential, number of blue
eyed eskimos) that I am not considering. What factors do you think I
should have used? which factors should I have not used? Should the
factors be weighted differently? These are all interesting questions
and I would love to hear your answers.
(2) If it were completely up to me, I would not modify the seedings
(3) It is a legitimate concern that the bracket could end up as a
replay of sectionals. this is undesirable from a competition and
fairness perspective
(4) Given the possibility to address (3) with minor changes to the
initial seeds, I think the change is worth considering and is less
offensive than other alternatives that I have seen.
-MH
the bama section is moot? thats an angry statement. sure none of the
bama section teams are C1 worthy, but i'd pick vandy or auburn to be
in regionals over UT-Arlington or Houston. dont underestimate the bama
section just because its small and doesn't get much national
attention. and to seed the winner of the section as 9th out of 16?
thats pretty bold! lsu doesn't deserve first or second, or even up to
the 4th. the fifth seed is a long stretch, but surely they're above
the 9th seed, in the top half at least. i implore the people to
reconsider the rankings.
hahahaha.
Be careful what you say about Houston. I hear that a certain skinny
fellow with sticky fingers playing with them this year.
-DW
That fast kid with glasses who sometimes whines about calls?
-CS
well i haven't seen houston play, so i'm not sure about their level of
play, i was just going by upa.
I watched UH v ATM and that dude was throwing pretty well in the
wind. However his sticky fingers must have changes since leaving Wash
U because he must have had 7 or 8 drops on dump passes. Pretty tough
to watch. Houston was surprisingly not bad.
oops, maybe i shouldn't rag on them then. either way they're in
regionals so they must be a decent team at the least.
they've been in and out of sectionals the past 5 years or so.
Sometimes tehy have a team, some times they dont.
allen
Wow Mike, way to stick up for your former Region.
Leave the Bama Section alone!
-MH
Some historical Regionals perspective:
2008: Bama sectionals teams enter at 6, 10, 13, 15 (average 11) and
leave at 9T, 11T, 11T, 13T (average 11)
2007: 3 of the 4 Bama sectionals teams broke seed in pool play and LSU
and Auburn finished above seed (I have no idea about Vandy and
Mississippi, as score reporter is incomplete and the 17 team regionals
throws off traditional snake seeding.) - LSU finishes top 3 (breaking
seed by at least 2)
2006: Bama section enters at 5, 9, 14 (average 9.333) and leave at 5,
7, 9 (average 7) - LSU finishes top 5
2005: Bama section enters 8, 9 and leave 5T, 9T (LSU not in attendance
due to DQ after sectionals.)
2004: Enter 5, 6, 11 and leave 2, 5T, 9T - LSU finishes top 2
(breaking seed by 3)
That's as far back as score reporter was cooperative. So, every year
that LSU has been to Regionals, they have held or broken seed,
finishing top 5. The section as a whole tends to break seed. The
bottom Bama teams are generally better than the bottom teams from
other sections and LSU is generally (read: always) better than over
half the field.
I have no idea how things should be seeded this year, as I have played
against practically 0 of these teams, but the Bama section in general
and LSU in particular traditionally get underseeded.
-McB
Ozark Section
Coming in seeded (3,4,5,8,11,12) average = 7.16
Placement (3,4,5T,7,8,14) average = 6.83
Texas Section
Coming in seeded (1,2,7,9,14,16) average = 8.16
Placement (1,2,5T,9T,15,16) average = 8
Looks like every section broke seed!!!
Those ties naturally bring down the average on the final results...
The other years are probably more legit, but that is still a bad way
to look at it.
There are also other factors that affect final placement, like last
year our team (Truman State) only took 9 guys down to Regionals
because it was an 14 hour drive and ended up playing savage.
Initially seeding it would have never factored in something like that.
Wow!!! Tied for 9th=9th!
LSU dominated last years regionals!
1) Texas (win over kansas)
2) Kansas
3) Arkansas (win over North Texas)
4) North Texas (superior season results than the next options Wash U
and LSU)
5) Texas State (win over Wash U, superior results to LSU)
6) Texas A&M (win over LSU, mixed results with Wash U but higher RRI)
7) Wash U
8) LSU
9) Truman
10) Mizzou
11) Ole Miss
12) Rice
13) Harding
14) Miss St
15) Houston
16) UTA
much of the bottom seeds based solely on RRI.
My opinion seems to be close to that of the original post, with the
main exception being Arkansas ahead of both North Texas and Texas
State due to arkansas' head to head over NTexas.
If you disagree with any positions, please give reason to support your
disagreement.
Your rationality isnt wanted here. The whole point that matters is
that we're dissing the Bama section and they never get respect!!!
It's all because the man wants to put the bama section in its place!
LSU didn't make Regionals last year.
Last year all the Bama teams were eliminated from contention after the
3rd round.
Bam. Someone just got Doeged.
ORLY! Way to pickup on the sarcasm
yea, this thread needs a Mad Dog comment
Teams
1. Texas 2543 TEX
2. Kansas 2591 OZ
3. Arkansas 2365 OZ
4. North Texas 2463 TEX
5. Texas State 2488 TEX
6. Wash University 2365 OZ
7. Texas A & M 2457 TEX
8. LSU 2364 BAMA
9. Truman State 2458 OZ
10. Missouri 2332 OZ
11. Rice 2315 TEX
12. Mississippi 2333 BAM
13. Harding 2287 OZ
14. Mississippi State 2169 BAM
15. Houston 2205 TEX
16. Texas-Arlington 2002 TEX
1:Texas
16:Texas-Arlington
9:Truman State
8:LSU
5:Texas State
12:Mississippi
13:Harding
4:North Texas
3:Arkansas
14:Mississippi State
11:Rice
6:Washington University
7:Texas A & M
10:Missouri
15:Houston
2:Kansas
Who is this? Docter?
I understand your argument, but hypothetically... Texas State is the 5-
seed, that puts them in a second round matchup against the 4-seed. If
you switch them, they have to play the 3-seed which you could argue is
a disadvantage they don't deserve, and Arkansas is also at a
disadvantage having to play the them, as opposed to the team that
deserves to be the 6-seed.
I completely understand what you are saying, but teams could be
considered at a disadvantage if you switch seeding based on matchups.
-casey
I would just like to see the teams in the semis play teams that would
not be from their own section.
D Rich
(that better Bigley)
I just don't really see how sectional affiliation should be a factor.
This isn't a tournament where regional diversity of competition should
play any role in scheduling (at least not a role that outweighs more
merit-based factors). If two Texan teams are deemed to be the 4th and
5th best teams in the region, they should be playing each other; it
shouldn't matter than they're both Texan teams. Like Casey said, if a
team is the 5th best team in the region (based on merit) and you make
them the sixth just for the sake of regional diversity, you are
harming the 3rd seed and helping the 4th seed by creating 3/5 and 4/6
matchups. At this point in the season, you have earned your seed at
regionals, and should have to play whoever else has earned their
corresponding seed. Especially when it gets to semis or finals, you
can't account for sectional affiliation.
Jack
(Don't worry D Rich, I hate Bigley too)
Good point. I guess that is why they have the back door game.
Well good luck to all teams and I wish I could be there to get drunk
and heckle with all the Alumn from rival schools.
D Rich
(Still trying to figure out which Grad school will accept me next year
so I can use my last year of eligibility)
Jack, you do well in stating this theory, but you miss a rather large
hole in the logic of it. You presuppose the existence of strong
evidence that any of the seeds 3 through 6 truly belong in that
position relative to the other teams. The evidence really just isn't
there, I wish it were, but we don't have anything outside of a meager
and incomplete head to head tally and the often unreliable common
opponents metric. Lacking that evidence, your rationale for not
swapping teams falls a bit flat.
It sounds like Kansas feels Arkansas is an easier opponent and is
scared of UNT.
I don't think anyone from Kansas is saying that. Especially with K-
rich apparently being out with due to his appendix (if this is in fact
true I feel awful for him, I would have loved to played against one of
the best the South has to offer, and hope he heals up fast).
-KK
I would suggest the West Coast. Weather is beautiful and so are the
girls (as long as you don't go to Stanford). Personally, I think it
is pretty awesome that sectional rivalries will be born out in full at
regionals. Good luck to everyone and much love from the NW.
-DW
(Now of Stanford)
P.S. - A little birdy told me that Houston might not be attending. 15
team regionals anybody?
Really? I was kinda looking forward to a re-match of the 2005 Forth
Worth Bowl
Actually, that's pretty much what I was saying. Scared to death of
the Texas section. You caught me.
Fuck seeds. Win your games. We are playing in our-kansas. There
better be cheap cigarettes there.
You know, I for sure thought when you moved out to Stanford you would
forget all about the plain ole' South Region and move on to the glitz
and glam of the NW. What is it the grass isn't greener over there?
SLAM Superfly! It sounds like there hasnt been any improvement over
WUWU's looks (jk WUWU, you guys are hot).
I'm pumped for potential second round matchups provided that teams
hold seed. The 6 v 3 game gives WashU a chance to avenge its 2 losses
to Arkansas at Sectionals (on a personal note, that 2nd loss was
absolutely brutal in such a physical and hard fought game). Texas A&M
can prove their win over Kansas at Centex was not a fluke in the 2 v 7
game. UNT and A&M can have a Texas style sectionals re-match, and
then there is TUFF vs LSU. But nobody cares because LSU is in the
'Bama section.
There's a Bama section? First I've heard of it.
-Joey
Of course, Stu, you would care more about the womens' seedings. Lets
not forget who your favorite team is to coach...
And yes Kate, TSUnami will always have a special place in my heart.
Did we lose a game between Centex and Nationals? Fool's Fest doesn't
count b/c I was with the Z alum. I did like your sideline advice for
the KU boys. I believe it went "Always listen to Stu!" Makes me look
infallible.
Rock Chalk Little Rock!
P.S. Thanks for the cigarette pricing breakdown Jacob.
Something about tournaments in forty degree rain and twenty mph winds
that brings a section closer together. Ozarks for life.
Did I mention win your games?
Win, or go home. We don't get paid to do this, fight for your pride!
sTupac
Dennis Tarasi
South Open RC
515-419-4110 (preferred)
isaratd at hotmail.com
I understand not wanting to make the trip for a double elimination
bracket, but I don't understand waiting until the last minute to give
your definite answer to the RC.
lame move houston, I'm sure a number of ozark teams would've been able
to make the trip.
Infact, I wish OU was going to be there (or whichever ozark team that
finished higher).
maybe there's still time to get a squad in, and maybe we'll just have
to play a 15 team regionals----
------rickner
tx state
when club regionals is in florida, I play sectionals knowing the
answer I'm giving the RC.
-Kevin
The regional coordinator's roommate
Kansas 15 - Miss St 5
Truman St 15 - LSU 12
Missouri 15 - Texas A&M 11