Please look @ my power ratings + give me any ideas on how to make it
better!

This is from Wayne Smith, who administers a mailing list for college
hockey.  While most peole on this list have primary interests in college
football, I thought this information might be interesting and/or useful.
Cheers!

      - Pat



-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...


---------- Forwarded message ----------
Date: Fri, 28 Feb 2003 15:30:59 EST
From: Wayne T. Smith <email withheld>
To: Hockey-L@...
Subject: Heal Point System Rank for Men's Division I College Hockey,
              through 2003-02-25

Heal Point System Rank for Men's Division I College Hockey, through
2003-02-25, comments on HEAL, as well as the RPI and RHEAL rankings, may
be found on the web at

       http://Maine.edu/HEAL/

You do *NOT* see HEAL rankings here, because I believe it's time for the
NCAA Men's D-I Ice Hockey Committee to shed its current mish-mash of
yearly tweaked, eclectic selection and seeding rules for the ultimately
fair and defendable system called KRACH.

Anyone that knows my interest in college hockey ranking systems knows
that I've believed KRACH to be the best system for some years now. It's
not as simple as HEAL, but it is better.

Many or most ranking system geeks, a classification to which I am only a
poor, infrequent pretender, know that KRACH is a sound, terrific system.
Moreover, it is substantially better than other college hockey ranking
systems, including current and past NCAA systems, and (sigh) HEAL and
RHEAL.

I invite you to view the current KRACH ranking at USCHO.com

    http://uscho.com/polls/?data=krach

A nice KRACH FAQ is available at

    http://uscho.com/FAQs/?data=krach

For more information on KRACH and college hockey ranking systems, see
this web site by John Whelan, a great friend of college hockey ranking
(highly recommended!):

    http://slack.net/~whelan/tbrw/tbrw.cgi?krach

And now, if you're only going to read one article, read this:

    http://uscho.com/news/2003/02/27_006294.php

the February 27, 2003 article by Adam Woden of USCHO, entitled "No More
Gimmicks ...  It's Time for KRACH".

So, yes, I'm jumping on the band wagon.  Please join in.  Make sure the
NCAA Men's D-I Ice Hockey Committee knows your view!

cheers, wayne

Wayne T. Smith
wts@...                          Old Town Landing
HEAL/RHEAL and RPI at                  http://maine.edu/heal/
The College Hockey lists: Hockey-L and Hockey3 at http://lists.Maine.edu/

The problem is that it's unfair to dictate how computer ratings authors
should write our systems when their is no way to enforce those dictums
upon the voters in AP and Coaches polls.  Not to mention, the authors
have no voice comparable to that of the writers or the coaches who are
routinely on TV and the radio.  About a week ago, we witnessed Kirk
Herbstreit, one of the best college football analysts out there, smash a
computer on the ground during his College Gameday show in a display of
his feelings about the computer systems.  This is a huge conflict of
interest considering that Herbstreit is one of the AP voters and
therefore is saying that only the writers know what's best.  I hate to
single him out, but this is the kind of thing that has been happening on
all of the media outlets with almost no retort from the computer system
proponents.  When this is all that people see, most have no other
information but to think that the computers are wrong -- and that's
wrong and biased in and of itself.

As far as the debate about what's more important, I believe it's more
important to have a set of ratings that people understand and can point
out why such and such teams are rated where they are.  This applies not
only to the computer ratings, but to the BCS and the pollsters.  Right
now, the BCS system is so convoluted that most people don't understand
how it comes to the conclusions that it does.  And since the votes of
the pollsters are anonymous, we have no recourse to question how they
voted, yet the computers are wide open to debate.  Overall, the system
really isn't that complicated, but it's beyond the attention span of
most people.  So what needs to happen?

First off, the BCS has to let the computers do thing their way.  You
can't tell one group to do it one way and another to do it however they
feel -- it's a double-standard and it's the reason many of last year's
computer systems pulled out of the BCS this year.  That means let them
determine the best way to rank.  And if they want to tell the world how
they got there, then that's the authors' prerogative.  It's no different
than the polls.

Second, we need to simplify the BCS by eliminating the frivolous bonuses
like SOS, QW's, loss counts, and other things that are already figured
into both the polls and the computers.  All it really does is give
double bonuses (or penalties) to teams that trigger them, and since
that's considered already, why double (or even triple) consider it?
Without it, we probably would have had Oklahoma-Miami in 2000,
Miami-Oregon in 2001, and what seems to be a more correct ranking of
Miami-OSU at this point in the current season instead of the other way
around.

But most of all -- playoffs.  Have the BCS system be the governing body
that selects the playoff members -- let's say the top 8 teams -- and it
finally becomes worthwhile.  The problem lies with the bowl
commissioners and the AD's of the schools, but throw enough money at a
solution and they'll listen.

Just my $0.02...

--Jeff
imes@...
www.comprank.com



On Thu, 2002-11-14 at 15:25, Patrick E. Fleming wrote: Welcome to our
new members since the last post to the list!

Here is a topic on which I am interested in peoples' oppinions.  It
seems
to me that the the court of public oppinion, computer rating systems get
low marks because people don't tliek the results.  My contention is tha
tlooking at the results is exactly the wrong way to judge computer
ratings
systems.  Rather, I prefer to look at the rating criteria - and decide
if
the criteria are fair.  Trying to adjust the criteria in order to get
better results seems to be completely backwards to me.  (Of course, this
is exactly what the BCS did this past summer when the eliminaterd margin
of victory from the computer data . . .presumably because they felt
oregon
got unduely screwed over . . . what they failed to consider was that is
was the average strength of schedule in the pac-10 tha tpulled oregon
down
. . but that's beside the point.)

What do you all think?  Are the results more important than the method?
What sorts of methods (criteria) are most important?

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...



To unsubscribe from this group, send an email to:

     CompRatings-unsubscribe@yahoogroups.com



Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

Welcome to our new members since the last post to the list!

Here is a topic on which I am interested in peoples' oppinions.  It seems
to me that the the court of public oppinion, computer rating systems get
low marks because people don't tliek the results.  My contention is tha
tlooking at the results is exactly the wrong way to judge computer ratings
systems.  Rather, I prefer to look at the rating criteria - and decide if
the criteria are fair.  Trying to adjust the criteria in order to get
better results seems to be completely backwards to me.  (Of course, this
is exactly what the BCS did this past summer when the eliminaterd margin
of victory from the computer data . . .presumably because they felt oregon
got unduely screwed over . . . what they failed to consider was that is
was the average strength of schedule in the pac-10 tha tpulled oregon down
. . but that's beside the point.)

What do you all think?  Are the results more important than the method?
What sorts of methods (criteria) are most important?

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

OFFENSIVE YDS/PT = TOTAL YARDS/NO. PTS SCORED (<10 IS CONSIDERED
GOOD)DEFENSIVE YDS/PT = TOTAL YARDS GIVEN UP/NO. PTS GIVEN UP (>15 IS
CONSIDERED GOOD)MY OFFENSIVE TOTAL YARDS is = Rushing Yards + Passing Yards +
Punt Return Yards + KickOff Return Yards - Penalty Yards - Sack YardsDefensive
total yards is more or less the same formula, except from the defensive side.My
problem is I have not figured out how to involve turnovers, strength of
schedule, yards/pass attempt and changing the home field advantage (I never
believed in just three points per game all year)Hope this helps--- On Wed 10/09,
nbnfba < epastoor@... > wrote:
From: nbnfba [mailto: epastoor@...]To: CompRatings@...: Wed,
09 Oct 2002 12:47:50 -0000Subject: [CompRatings] Ratings formulasHello all. I
just found this list, and this seems like a really interesting list. I have a
question to start with. I currently run a fantasy football league where i am
developing a power rankings system that i would like to be something like the
BCS. Currently i have developed 3 formulas for the system. However, I would like
to have somewhere around 10 formulas. After three formulas though, i am running
out of ideas in my head on how to create new formulas. I was wondering if any of
you have any of your own formulas you wouldnt mind sharing, or at least posting
some version of pseudo-code, or a website that might have some suggestions about
creating more formulas. Obviously i would like to have as many formulas as
possible. My rankings have changed a lot as i continue to add more formulas.
Thanks for your help!~eric pastoor------------------------ Yahoo! Groups Sponsor
---------------------~-->Plan to Sell a
Home?http://us.click.yahoo.com/J2SnNA/y.lEAA/MVfIAA/jacolB/TM-------------------\
--------------------------------------------------~->To unsubscribe from this
group, send an email to:CompRatings-unsubscribe@... use of
Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

------------------------------------------------
Can a Web portal forever change your life?
Win up to $25 Million on iWon - click here!


[Non-text portions of this message have been removed]

Hello all. I just found this list, and this seems like a really
interesting list. I have a question to start with. I currently run a
fantasy football league where i am developing a power rankings system
that i would like to be something like the BCS.  Currently i have
developed 3 formulas for the system. However, I would like to have
somewhere around 10 formulas.  After three formulas though, i am
running out of ideas in my head on how to create new formulas. I was
wondering if any of you have any of your own formulas you wouldnt
mind sharing, or at least posting some version of pseudo-code, or a
website that might have some suggestions about creating more
formulas. Obviously i would like to have as many formulas as
possible. My rankings have changed a lot as i continue to add more
formulas.
Thanks for your help!

~eric pastoor

Does anybody here have any experience using game output functions.

I'm working on a ranking system for baseball and every time I use the
margin of victory, the Boston Red Sox come out on top.  The have the
largest margin of victory of any team even controlling for their
strength of schedule.  I would rather not use a cap.

Any help would be greatly appreciated.


Alan

I will publish a preseson ranking of college football teams based on
last years finish and this years schedule.  This is a new method I will
use this season.  I have a home advantage factor in my rankings and I
consider this very important.  I use this as a baseline until at least
the middle of October.  I have used different methods in previous
seasons and none have worked very well for me.

My system always starts the teams equal but I'll use a baseline for
some type of history until enough data is collected to turn it loose.
I'm always interested in improvement so I'll see how this method
developes.

Regards,

Randy Mac

--- "Patrick E. Fleming" <pfleming@...> wrote:
>
> Okay, the first preseason poll is out and can be seen (among other
> places)
> at
>
>      http://sports.yahoo.com/ncaaf/rankings/coachpoll.html
>
> So, the question is, how do people rate teams for computer systems
> before
> any games are played?  In my system, all teams are rated equally
> before
> the season begins.  But that is only one way of doing things.  How do
> the
> rest of you do it?
>
>      - Pat
>
> -- Patrick E. Fleming
> -- Department of Chemistry
> -- San Jose State University
> -- pfleming@...
>
>
>


__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

I guess I'll put my two cents worth in this discussion.  I cannot think
the BCS actually believes they will have a better system because the
MOV (Margin of Victory) is eliminated.  The continuing change of their
formula is not because of computer ratings but is a reflection of the
quality of teams playing during the respective season.

In my opinion and my ratings, Oklahoma was very superior to anyone else
during the 2000 season and Miami was far and beyond the best team in
2001.  The cry about who was going to play them is nothing new.  During
seasons when the top teams were fairly equal the cry went up about who
was going to play in the other six spots.  The BCS is all about money
and not truly dedicated to determining a National Champion.

I wonder how many sports writers and coaches consider the MOV when
their ballots are cast.  I would bet if the truth were known all of
them do yet the BCS has not asked the persons who vote in polls to
change their way of thinking.

The computer rankings which have changed their formulas to satisfy the
BCS are flawed in their development.  The persons who have not changed
their formulas due to pressure have earned my deepest respect.

I think any computer poll which never considered the MOV should have
been used by the BCS instead of forcing people to change their
rankings.

  My opinion in determining a National Champion remains to have at least
30 computer rankings (with different methods of determining the rating)
counted as one poll with the AP and Coaches Polls counting as one poll
each.  This would give us a consensus opinion regarding which team is
best.

Regards,

Randy Mac

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

I'm working on baseball right now.  I'm starting
off with a model of wins and losses.  I have found
another problem with modeling wins and losses.
Teams that haven't won or lost a game yet are difficult to rate especially
with logistic regression.

Colley's Matrix has a nice little adjustment
courtesy of Laplace that alleviates that problem, but the model itself can't
handle homefield advantage or change over time.

In baseball you get this problem with starting pitchers.  Albie Lopez started
four games for the
Braves and the Braves lost all four games.  I tried merging his games in with
the two games started by AAA callups but they lost their two games also.

In baseball it seems to help if you take the square-root of the runs scored by
each team and then do a regression for runs scored.  It reduces
heteroscedasticity and keeps runaway scores from skewing the estimates.

The thing I like about modeling wins directly is that you get an estimate for
one team beating another.  Modeling margin of victory, you get a winner, but
usually you have to do some simulation to get an estimate of the probability
of winning.

Alan Jordan


On Sun, 4 Aug 2002 04:49:04 -0400 Brad Kiser <rkiser@...> wrote:

> I do use Margin of Victory. I am of the opinion
> that ignoring MOV is ignoring
> a very informative chunk of data, which is a
> statistical no-no. My MOV is
> capped, so SOS is more important to a team's
> rating, but it can make the
> difference between two similar teams.
>
> Who here does/doesn't use MOV?
>
> - Brad
>

I do use Margin of Victory. I am of the opinion that ignoring MOV is ignoring
a very informative chunk of data, which is a statistical no-no. My MOV is
capped, so SOS is more important to a team's rating, but it can make the
difference between two similar teams.

Who here does/doesn't use MOV?

- Brad


>===== Original Message From alandjordan2002 <ajordan@...> =====
>--- In CompRatings@y..., Brad Kiser <rkiser@u...> wrote:
>> I don't rank at all until I have enough data, which is probably
>going to be
>> late September. So my teams are rated strictly on performance this
>season
>> regardless of who they are or what they did last year.
>>
>> -Brad Kiser
>>
>
>
>Do you factor in Margin of Victory in your ratings or do you use
>just wins and losses?
>
>Alan Jordan
>
>
>To unsubscribe from this group, send an email to:
>
>    CompRatings-unsubscribe@yahoogroups.com
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

--- In CompRatings@y..., Brad Kiser <rkiser@u...> wrote:
> I don't rank at all until I have enough data, which is probably
going to be
> late September. So my teams are rated strictly on performance this
season
> regardless of who they are or what they did last year.
>
> -Brad Kiser
>


Do you factor in Margin of Victory in your ratings or do you use
just wins and losses?

Alan Jordan

Thank you.  That almost makes sense.

Alan Jordan




On Sat, 3 Aug 2002 01:35:33 -0700 (PDT) "Patrick E. Fleming"
<pfleming@...> wrote:
>
> I can help you on this point.  (microsoft
> products are oh-so-intuitive . .)
>
> To get the inverse of a matrix in excel, first
> enter the matrix in an n by
> n square set of cells.  Next, highlight the n
> by n square set of cells in
> which you want to place the inverse of the
> matrix.  In the upper left hand
> cell, type (but don't hit return yet!):
>
>      =minverse(b3:d5)
>
> (assuming the original matrix is contained in
> the group of cells with
> upper left corner being cell b3 and the lower
> right corner beig cell d5).
> Now, rather than simply hitting return (which
> will give you the upper left
> hand member of the inverse matrix only) you
> mist use cntl-shift-enter.
> (See? I told you it was intuitive!)
>
> You can check your result by multiplying the
> inverted matrix by the
> original.  Highlight an n by n space where the
> product matrix will go, and
> enter in the upper lefthand corner cell (again
> not hitting return yet!):
>
>      =mmult(b3:d5,f3:h5)
>
> assuming the original matrix is in the space
> b3:d5 and the inverse is in
> the space f3:h5 as defined above.  Now hit
> cntl-shift-enter and you should
> get an identity matrix.
>
> Hope this helps!
>
>      - Pat
>
>
>
> -- Patrick E. Fleming
> -- Department of Chemistry
> -- San Jose State University
> -- pfleming@...
>
>
>
> To unsubscribe from this group, send an email
> to:
>
>     CompRatings-unsubscribe@yahoogroups.com
>
>
>
> Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/
>
>
>

Alan wrote:

> Correction - Excel is supposed to do an inverse of a matrix but I couldn't get
> minverse to work right.  It kept giving me a scalar (single value) instead of
> a matrix.  I ended up having to run the matrix part of the example in SAS.

I can help you on this point.  (microsoft products are oh-so-intuitive . .)

To get the inverse of a matrix in excel, first enter the matrix in an n by
n square set of cells.  Next, highlight the n by n square set of cells in
which you want to place the inverse of the matrix.  In the upper left hand
cell, type (but don't hit return yet!):

      =minverse(b3:d5)

(assuming the original matrix is contained in the group of cells with
upper left corner being cell b3 and the lower right corner beig cell d5).
Now, rather than simply hitting return (which will give you the upper left
hand member of the inverse matrix only) you mist use cntl-shift-enter.
(See? I told you it was intuitive!)

You can check your result by multiplying the inverted matrix by the
original.  Highlight an n by n space where the product matrix will go, and
enter in the upper lefthand corner cell (again not hitting return yet!):

      =mmult(b3:d5,f3:h5)

assuming the original matrix is in the space b3:d5 and the inverse is in
the space f3:h5 as defined above.  Now hit cntl-shift-enter and you should
get an identity matrix.

Hope this helps!

      - Pat



-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

> Can you say a few more words about the Colley
> Matrix?  What is it? What
> does it do?  Are there any references you might
> suggest for further
> information?  Thanks and cheers!
>
>      - Pat
>

The Colley Matrix is one of the components of
the BCS system.  It is designed and run by Wesly N. Colley and is printed in
the Atlanta
Journal and Constitution.

It is from the Just Win Baby (JWB) school of thought and does not factor in
margin of victory
(MOV).  It also does not factor in home field advantage or most recent games.
Everyone starts the year off equal.

What it does do is control for the strength of each individual opponent the
team has played.
The elegant part about it is that once the data are in order it calculates the
rankings without having to iterate the way maximum likelihood or nonlinear
least squares would.  All it requires is inverting a matrix and muliplying it
by a vector.  That's the beauty of it.

The other great thing about this system is that the author tells you exactly
how he does it.  There are no hidden fudge factors.  I hate it when people
refuse to divulge how they do it.



You can find out more about it at his web site including an example.
http://www.colleyrankings.com/




Correction - Excel is supposed to do an inverse of a matrix but I couldn't get
minverse to work right.  It kept giving me a scalar (single value) instead of
a matrix.  I ended up having to run the matrix part of the example in SAS.

Alan

alandjordan2002 wrote:

> I think you should make the out of conference games a little more
> rigorous.  I could show you a way based on the way that the colley
> matrix does it.  It requires some data organization, but once its done
> you can do all the calculations on Excel.

Can you say a few more words about the Colley Matrix?  What is it? What
does it do?  Are there any references you might suggest for further
information?  Thanks and cheers!

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

I think it's rather pointless to do a preseason and early season ranking like
some do. Sure, you can use data from last year, but exactly what purpose does
that serve? Unless your system is predictive, it is useless to do ratings
until you have enough data to run this season's games only.

I don't rank at all until I have enough data, which is probably going to be
late September. So my teams are rated strictly on performance this season
regardless of who they are or what they did last year.

-Brad Kiser







>===== Original Message From "Patrick E. Fleming" <pfleming@...> =====
>Okay, the first preseason poll is out and can be seen (among other places)
>at
>
>     http://sports.yahoo.com/ncaaf/rankings/coachpoll.html
>
>So, the question is, how do people rate teams for computer systems before
>any games are played?  In my system, all teams are rated equally before
>the season begins.  But that is only one way of doing things.  How do the
>rest of you do it?
>
>     - Pat
>
>-- Patrick E. Fleming
>-- Department of Chemistry
>-- San Jose State University
>-- pfleming@...
>
>
>
>
>To unsubscribe from this group, send an email to:
>
>    CompRatings-unsubscribe@yahoogroups.com
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

--- In CompRatings@y..., "pete42869" <pete42869@a...> wrote:
> I would welcome any feedback on this;


I think you should make the out of conference games a little more
rigorous.  I could show you a way based on the way that the colley
matrix does it.  It requires some data organization, but once its done
you can do all the calculations on Excel.

The idea is to collapse the out of conference games as if each
conference was a single team.  Then run them through the Colley
Matrix.  There are other ways that are more advanced, but require
SAS, SPSS or some major programming.

Alan Jordan

Beginning last year, I took the previous season's ratings and used that
as a seed for the few 4-5 games, similar to what Jeff Sagarin does.  The
actual games are weighted much more heavily than the initial rating, so
corrections are made quickly, but it tends to work well.

I'm working on a new method this year which takes into account returning
starters and possibly coaches since that seems to be a relevant
statistic, but I'm not sure how it will pan out.  We'll see.

--Jeff

On Thu, 2002-08-01 at 19:09, Patrick E. Fleming wrote:
>
> Okay, the first preseason poll is out and can be seen (among other places)
> at
>
>      http://sports.yahoo.com/ncaaf/rankings/coachpoll.html
>
> So, the question is, how do people rate teams for computer systems before
> any games are played?  In my system, all teams are rated equally before
> the season begins.  But that is only one way of doing things.  How do the
> rest of you do it?
>
>      - Pat
>
> -- Patrick E. Fleming
> -- Department of Chemistry
> -- San Jose State University
> -- pfleming@...
>
>
>
>
> To unsubscribe from this group, send an email to:
>
>     CompRatings-unsubscribe@yahoogroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>

Okay, the first preseason poll is out and can be seen (among other places)
at

      http://sports.yahoo.com/ncaaf/rankings/coachpoll.html

So, the question is, how do people rate teams for computer systems before
any games are played?  In my system, all teams are rated equally before
the season begins.  But that is only one way of doing things.  How do the
rest of you do it?

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

Pat, et al,

I'm glad someone finally put down in words what I've been thinking about
all off-season -- that is if the teams had been normalized based on
their relative rankings instead of their ordinal values, that we may
have had a different contender against Miami in the Rose Bowl last
January.  The BCS officials don't seem to understand this problem, but I
think you're right in that they really don't seem to care.

On my website last year, I wrote a small piece on the use of the BCS in
a college football playoff system and mentioned that the way that the
BCS is setup is not to guarantee that the best two teams will matchup in
the BCS championship, but two highly qualified teams.  In many years,
there are truly only two great teams (1994, 1997), sometimes there's
only one (2001), sometimes there's several (1968, 1984, 1990).  The neat
thing about the old system is that there may have been more than one
ultimate bowl game to watch so it generated more interest.  The bad
thing is sometimes, that didn't solve anything.  The new system takes
that interest out though, without giving the next best teams the chance
to prove themselves.  In addition, pitting what some feel to be an
unworthy opponent in a winner-takes-all game can have huge
repercussion.  Had Nebraska actually beaten Miami last January, the
whole system could have fallen apart.  A playoff system would correct
these problems.

As for this year's changes, I don't see it as being neutral, but instead
it will likely be less consistent due to the fewer amount of computers,
more convoluted to the general public, and favors teams that play weaker
schedules now due to the elimination of MOV.  The BCS is toying with
things that are already accounted for in many computer programs, double
counting others like SoS, and generally making a mockery of the whole
process.  The problem, IMHO, is that either they just don't get it, or
are intentionally trying to cause controversy where none needs to
exist.  I would not be surprised if there were 4-5 teams that "qualify"
for the BCS title game this year.

The bottom line is that I believe that a playoff is necessary, and not
unattainable with the current system.  I urge people to go and read my
little article at http://www.comprank.com/opinions.html#bcs and comment
if you like.  I'm not sure that my idea is a silver bullet solution, but
it has got to be better than what we have now.

In any case, looking forward to 2002 college football!  Only 24 more
days!

Jeff Imes
http://www.comprank.com

I suggested earlier the changes to the BCS as a topic for discussion.  So
here are my two cents.  I think the changes are completely neutral in
terms of what they will fix.  I think that this is true for a couple of
reasons.

First, the question of whether to include a margin of victory or not is
not a simple black and white issue.  I think it is useful to include a
variety of ratings systems that do things in many different ways.  This
is, after all, the whole point behind having more than one person vote in
the polls.  The thing is that defining the "quality" of a football team by
determining a single numerical index is an impossible task.  A football
team is too complex a system.  Some teams are good because of offense,
some because of defense, some because of coaching, etc.  Not all systems
will treat these criteria equally so naturally, some teams will do better
than others in certain rating systems because the teams' particular
strengths are being rewarded disproportionately in a specific system.
For that reason, it only makes sense to have a variety of different
systems included.

That said, I think the BCS is screwing things up unnecessarily because
they don't seem to be able (interested?) to do simple math.  In both the
polls and the computer ratings, there is a measure not only of ranking,
but also of how close one team is to another.  Using only the ordinal
ranking and ignoring the actual average rating (or total points in the
polls) is simply unnecessary and distorts the final results.  For example,
the top team could be three touchdowns a game better than the second place
team, but by using only the ordinal ranking, the top team is exactly one
point better than the second place team.  Likewise, the second place team
could be a thousandth of a point "Better" than the third place team - and
yet the ordinals make #2 exactly one point better than #3.  Why?  This is
college football for crying out loud - surely the BCS can find someone who
understands statistics!  For a better and more in-depth discussion of
these problems, see:

      http://www.angelfire.com/ne2/bcs_exposed/

But finally, I don't think that any changes to the system will prevent
people from complaining.  The third place finisher will always claim they
deserve a spot in the championship game.  But then again, a playoff system
won't help either.  The ninth place finisher will complain in an
eight-team playoff, so what's the difference?  The only objection I have
to the BCS is how teams are chosen for the non-championship bowl games.
For example, if BYU won out last year, they should have gone to a BCS
Bowl. Marshall and Tulane should have had more consideration a couple of
years ago.  Sure - they don't play in the power conferences, but they
still managed to beat everyone who agreed to play them.  What more can you
ask of a team?

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

I would welcome any feedback on this;

I'm a college football fan and follow the various ratings fairly
close, but much more interested in a specific topic every year.

I've always been gripped by the annual debate about which is the
toughest conference, and always get disgusted because I rarely see an
unbiased opinion on this.

Keeping in mind that I am no math mathematician (I'm in the process
of trying to pass prob & stats for the 2nd time), and my relatively
low-budget cheesy formula for determining this is not mathematically
oriented, I would like for some of you (as college football fans more
than anything else) to take a look at my site & give me an honest
opinion.

I also run a NCAA Tournament website at www.tournamentfacts.com, full
of stats & trivia more than anything else.  My college football
conference ratings URL is www.angelfire.com/sports/conferenceratings -
Basically, I am just curious if anyone considers this a good
barometer to use (without the use of mathematical power ratings &
such), or is it just plain cheesy/hokey & a waste of my time.

I am considering purchasing a domain name & moving it from my free
server.  Any comments are appreciated.

It's a very bad idea, IMHO.  Any statistician will tell you that
having more data is preferable to less, and what they've effectively
done is limit the amount of data that they can use.  Most systems,
including mine, put caps on the MOV and that works well to counter
against RUTS'ing.  To ignore it altogether is basically saying that it
really only matters who you play and if you win.  That's far too
little to consider.

For example, if Team A and Team B play the same schedule, but don't
play each other, then if they both beat all the same teams, it
wouldn't matter if Team A won every game by 40 points and Team B won
by one point -- in the eyes of the BCS, they're identical quality
teams.  Asinine.

Their other big decision to tweak the BCS formula by only throwing out
the worst score of the computers is a step in the wrong direction too.
  At least be consistent with statistics and throw out both the best
and the worst, but now they have a system that can be skewed by a
system that ranks teams too high.  Basically, they didn't learn
anything over the past few years -- terribly unfortunate for college
football.

One thing that I really don't appreciate it the strong-arm tactics of
the BCS to make the computers do something the authors never intended
for their systems.  You can't enforce that on the writers and the
coaches, so why make the computers do it?  Here's the bottom line --
if you don't like the way a computer does things, then don't use it,
but don't make us create intentionally imperfect systems just because
you don't like it.  Find some other system to do your bidding -- I'll
keep doing it my way, which I feel is pretty accurate now after years
of work.

[steps off soapbox...]

Later,
Jeff
www.comprank.com


--- In CompRatings@y..., "Patrick E. Fleming" <pfleming@s...> wrote:
>
> As one suggestion for a topic of discussion, what are people's
feelings on
> the BCS descision to drop margin of victory from the computer
ratings
> portion?  Good thing? Bad thing? neutral?

To me it's a terrible thing. If the idea is to get the best estimate for how
strong a team is, how can you just throw out how badly they beat other teams?

I heard a guy call in to a local sports talk show a few weeks ago, discussing
this topic. He said when he devised his system, he simulated with a wide
variety of parameters, and to judge how well it did, looked at how well it
retrodicted the games from that year. According to these tests, the most
accurate way to rank the teams was to have no cap at all.

I personally have a cap on mine, but I use a special formula anyway to
consider other MOV factors.

As a statistician, my opinion is that it is asinine to throw away valuable
information like that. Their reasoning is terrible too, thinking that it will
keep coaches from running up the score. We all know that MOV has a larger
effect on poll voters than it ever does on computers anyway.

-Brad














>===== Original Message From "Patrick E. Fleming" <pfleming@...> =====
>Robert B. Kiser wrote:
>
>> Hello all,
>
>   Hi Robert and welcome.  This will be the first full college football
>season for the CompRatings mailing list.  I hope we can have some lively
>and insightful discussion (emphasis on the latter of course.)  I hope
>everyoen will feel encouraged to follow your lead and introduce themselves
>and  maybe share a little bit about their opinions, philosophies,
>mathematics and any other spects of their ratings systems.
>
>As one suggestion for a topic of discussion, what are people's feelings on
>the BCS descision to drop margin of victory from the computer ratings
>portion?  Good thing? Bad thing? neutral?
>
>I hope everyone is having a safe and enjoyable summer.  cheers!
>
>     - Pat
>
>-- Patrick E. Fleming
>-- Department of Chemistry
>-- San Jose State University
>-- pfleming@...
>
>
>
>
>To unsubscribe from this group, send an email to:
>
>    CompRatings-unsubscribe@yahoogroups.com
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

Robert B. Kiser wrote:

> Hello all,

    Hi Robert and welcome.  This will be the first full college football
season for the CompRatings mailing list.  I hope we can have some lively
and insightful discussion (emphasis on the latter of course.)  I hope
everyoen will feel encouraged to follow your lead and introduce themselves
and  maybe share a little bit about their opinions, philosophies,
mathematics and any other spects of their ratings systems.

As one suggestion for a topic of discussion, what are people's feelings on
the BCS descision to drop margin of victory from the computer ratings
portion?  Good thing? Bad thing? neutral?

I hope everyone is having a safe and enjoyable summer.  cheers!

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

Hello all,

I've just finished my ranking system and have put up a very crude website with
the results I have so far. You can check out my rankings for the 1996-2001
seasons at

http://web.utk.edu/~rkiser/kiser-gh6/index.html

I look forward to discussing the season and the rankings with you all.

-Brad Kiser

The BCS system itself is to blame for the choice of Nebraska to play Miami
in the Rose Bowl, not the computers.  Before the bowl games were played,
you could make the case that any of Oregon, Colorado, Florida, Tennessee,
Texas, Maryland, LSU, and perhaps even Nebraska, were all on or near the
same level.  To arbitrarily select just one of these teams is asking for
controversy.  A playoff to determine who gets to play in the MNC would have
been the only acceptable method of selecting the two best teams for the
championship game.

A few weeks ago, I published a short essay about this on my website
regarding how the BCS and a playoff can co-exist.  It may not be the
perfect solution, but it's better than what we have now, addresses the
issue of a playoff, and will allow the BCS to continue to play an important
role in determining who qualifies for the BCS games.  You can check it out
at http://www.comprank.com/opinions.html#bcs

Thanks,
Jeff

At 01:10 AM 1/5/02 -0800, you wrote:

>Okay - the 2001 season is over inc ollege football.  Miami managed to
>crush Nebraska in the Rose Bowl.  Given how the other BCS bowl games went
>(namely the number of blowouts) might it have made more sense for Oregon
>to be the Rose Bowl opponent of Miami?
>
>Given that it was the computer polls that managed to give Nebraska the nod
>over Oregon, what does this all mean?  Should there be changes? and if so,
>what?
>
>      - Pat
>
>-- Patrick E. Fleming
>-- Department of Chemistry
>-- San Jose State University
>-- pfleming@...
>
>
>
>
>To unsubscribe from this group, send an email to:
>
>     CompRatings-unsubscribe@yahoogroups.com
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

Okay - the 2001 season is over inc ollege football.  Miami managed to
crush Nebraska in the Rose Bowl.  Given how the other BCS bowl games went
(namely the number of blowouts) might it have made more sense for Oregon
to be the Rose Bowl opponent of Miami?

Given that it was the computer polls that managed to give Nebraska the nod
over Oregon, what does this all mean?  Should there be changes? and if so,
what?

      - Pat

-- Patrick E. Fleming
-- Department of Chemistry
-- San Jose State University
-- pfleming@...

Enter your vote today!  A new poll has been created for the
CompRatings group:

Should BYU be included in the Bowl
Championship Series?

   o yes
   o no


To vote, please visit the following web page:

http://groups.yahoo.com/group/CompRatings/polls

Note: Please do not reply to this message. Poll votes are
not collected via email. To vote, you must go to the Yahoo! Groups
web site listed above.

Thanks!