Coaches/Computer Rating
In order to consider factors which go beyond game scores, LaxPower has
adopted the Bowl Championship Series (BCS) formula for college lacrosse,
e.g., Coaches/Computer Rating (CCR) and a modified (to be determined) CCR
formula for high school lacrosse. There are five factors or components in
the current BCS formula:
(1) polls
(2) computer ratings
(3) strength of schedule
(4) quality wins factor
(5) loss factor.
This formula will be applied individually to College Men: Division I, II,
III, JUCO and USLIA and individually to College Women: Division I, II and III.
As an example as of April 23, 2001, Princeton topped the list with the lowest
score of 1.7. They received 1 point for ranking number one in the polls,
one point for ranking number one in the power ratings, 0.64 points for ranking
16th in strength of schedule (16/25 = 0.64), reduced this total by earning
1.90 quality win points and one point for one loss. The SOS was based on the
W-L percentage of their opponents and the W-L record of their opponent's
Opponent. The Quality Win Factor was based on the Polls (STX/USLIA) as opposed
to the actual BCS rankings. Massachusetts which did not receive a bit for the
tournament ranked 7th with a poll ranking of eigth, a computer ranking of fifth,
a strength of schedule twenty-third, 0.40 quality win points and one loss.
No Team CCR = Poll + PR + SOS - QWF + Loss
1 Princeton 1.7 1.0 1.0 0.64 1.90 1.00
2 Syracuse 4.2 2.0 2.0 0.32 2.10 2.00
3 Georgetown 7.4 5.0 3.0 0.40 2.00 1.00
4 Johns Hopkins 9.3 4.0 4.0 0.08 1.80 3.00
5 Notre Dame 9.8 3.0 6.0 1.08 1.30 1.00
6 Maryland 12.9 6.0 8.0 0.56 3.70 2.00
7 Massachusetts 14.5 8.0 5.0 0.92 0.40 1.00
8 Loyola 19.0 7.0 12.0 0.20 2.20 2.00
9 Virginia 24.2 12.0 9.0 0.04 2.80 6.00
10 Navy 24.3 13.0 7.0 0.44 1.10 5.00
11 Towson 28.8 10.0 17.0 0.84 1.00 2.00
12 Cornell 29.0 14.0 10.0 0.96 0.00 4.00
13 Bucknell 29.2 11.0 14.0 1.20 0.00 3.00
14 UMBC 33.6 16.0 13.0 0.12 0.50 5.00
15 Penn State 34.0 17.0 11.0 0.16 0.20 6.00
16 Hofstra 38.0 15.0 19.0 0.28 1.30 5.00
17 Duke 42.1 9.0 30.0 0.48 1.40 4.00
18 Pennsylvania 42.1 20.0 16.0 0.60 0.50 6.00
19 Army 44.2 19.0 18.0 1.16 0.00 6.00
20 North Carolina 58.5 18.0 35.0 0.24 0.70 6.00
For high school, the final ccr for 2001 is shownb below.
No Team CCR = Poll + PR + SOS - QWF + Loss
1 Landon -2.2 1.0 1.0 0.04 5.29 1.00
2 Loyola 7.1 4.0 2.0 0.36 3.27 4.00
3 Garden City 9.6 6.0 3.0 1.32 1.69 1.00
4 McDonogh 15.1 7.0 6.0 0.72 3.58 5.00
5 Yorktown 17.2 8.0 8.0 1.28 2.07 2.00
6 Gilman 25.0 5.0 19.0 0.56 3.58 4.00
7 Georgetown Prep 26.5 9.0 16.0 0.32 1.87 3.00
8 West Genesee 29.2 15.0 11.0 1.36 2.18 4.00
9 Northport 30.0 2.0 24.0 4.12 2.13 2.00
10 Boys' Latin 31.1 11.0 17.0 0.52 2.47 5.00
11 Taft 33.7 12.0 4.0 18.36 0.64 0.00
12 DeMatha 36.1 10.0 20.0 3.12 1.04 4.00
13 Brother Rice 36.4 14.0 7.0 15.24 0.87 1.00
14 Upper Arlington 38.8 27.0 9.0 1.84 1.02 2.00
15 Ward Melville 39.1 16.0 15.0 5.68 0.62 3.00
16 Wantagh 45.8 3.0 41.0 2.76 1.93 1.00
17 Moorestown 48.8 19.0 26.0 1.60 0.80 3.00
18 Sachem 52.8 18.0 25.0 7.44 1.64 4.00
19 Avon Old Farms 53.5 28.0 10.0 12.72 0.24 3.00
20 Loomis Chaffee 56.8 35.0 12.0 5.76 0.00 4.00
21 Liverpool 60.6 41.0 14.0 2.12 1.53 5.00
22 St Anthony's 66.0 21.0 22.0 21.36 0.33 2.00
23 Haverford School 66.9 24.0 37.0 0.92 0.00 5.00
24 Pinkerton 68.2 43.0 5.0 20.20 0.00 0.00
25 Fayettev'le-Manlius 70.5 25.0 40.0 3.64 1.11 3.00
Problems with this formula for high school include: (1) only 20-45 out of
teams are included in the polls. What number do we designate for the
remaining 1550+ teams in the CCR. One solution would be to treat teams
that receive poll votes as bonus points (instead of a rating of 1,2,3,etc.)
where the number of bonus points is given to teams with the greatest
votes. A second problem is "running up the score" creating a slightly
higher power rating. A solution to this is to have a 6th component which
is treated like the loss factor where after a certain goal margin for
games, a reduction number (like losses) could be implemented where this
number is proportional to the number of games and goals over a threshold
goal margin. This component is not a punitive measure but rather a
compensating factor to offset unreasonable gains. Another problem with
computer ratings is that game scores are not "normalized" for when the
game is played. Thus early season wins or losses count as much as late
season wins or losses and that includes the championship games. A solution
here is add a seventh component where we simply add all victories by a team
(to offset the the loss factor) and we adjust the victory and loss
components by date so that wins/losses occurring earlier in the season
count less than wins/losses occurring later in the season. Post season
tournament and championship games would then be more significant.
A modified (TBD) CCR will be used for high school lacrosse where the polls,
strength of schedule and quality wins factor will not follow the BCS proc-
edure exactly. In addition, a sixth and sevength component "RUST" or running
up the score component will offset gains made in the computer ratings because of
un unusally high goal margin (see above) and victories will further
emphasizing winning. The high school CCR will apply nationally to both boy's
and girl's high school varsity and club teams. The modifications to the BCS
formula will be based on inputs from boy's high school coaches.
(1) polls (treat as bonus points like quality wins)
(2) computer ratings
(3) strength of schedule (modified from BCS)
(4) quality wins factor (modified from BCS) influenced by date
(5) loss factor (influenced by date)
(6) running up the score component (not included in BCS)
(7) victory factor (influenced by date)
What Games Count?
For college, all games including post-season tournament games count.
For high school, all games appearing on a teams schedule will count as
official games and used in the ratings. All scrimmages or tournament
games occuring during the season will not count unless both coaches agree
prior to playing the game that the game(s) will count. (Contact:
baldo@schore.com). All post-season championship tournament games will
be treated as official games and will count.
Polls
College: For Men's Div I, II and III, the Poll will be STX/USLIA.
For Women, the poll will be Womens Division I (II and III) IWCLA Poll.
For USLIA, the poll will be the USLIA Top 25 Poll. For JUCO lacrosse,
the poll will be the NJCAA Men's Lacrosse Poll. Other polls may also be used
next year.
High School The high school poll is based on a poll conducted by
Laxpower and voted on by high school boys coaches. We currently do not have
a girl's high school lacrosse poll.
See Polls for more details.
Computer Ratings
The computer ratings will consider only this year's scores and not past year's
performance. At the beginning of the season, all teams are equal. After the
fifth week of the season, computer ratings are added the the CCR formula. The
power ratings are calculated so that if you subtract the power ratings of the
two opposing teams and add the home field advantage, the result should equal
the goal margin or difference in scores between the two teams. The power
rating is based on two components: a local component considering games against
"local" teams and an external component, based on games played against all
other teams. The sum of the two yields the overall power rating. See
analysis for more details.
Strength of Schedule
Strength of Schedule for college is based on the BCS formula = won/loss
percentage of an opponent and the won/loss percentage of the opponent's
opponents. The average is based on 2/3 of the opponents won/loss percentage
and 1/3 of the opponent's opponent won/loss percentage. The average of the
opponent's power ratings (used last year) is shown for comparison but
is no longer used in the college formula. The SOS ranking is divided by 25 to
determine the "quartile" for the CCR forumla. For high school, it has yet to
be determined whether the BCS SOS formula or an average or weighted average
of the power ratings of the opponent's will determine the the Strength of
Schedule. See SOS for more details.
Quality Win Factor
The Quality Win Factor is based on the number of wins against quality
opponent's. A quality opponent is defined as a team which is ranked in
the top 15 of BCS. We will use the top 15 teams for Men and Women divisions
I and III, USLIA divisions. For Men and Women division II and JUCO, LaxPower
will use the top 10 teams. A victory against the number 1 ranked team in the
BCS standing (polls) is worth 1.5 points and a victory against the 15th
ranked team is worth 0.1 points. For division II and Juco, a victory against
the top team will be worth 1.0 point and a victory against the tenth team is
worth 0.1 point. For high school, we may use the top 50 or 100 teams to
determine quality wins and this will be determined later. QWF for
more details.
Loss Factor
The Loss Factor is simply the sum of all losses by a team. This number
will be the same for both the national and local ratings. A time factor
can be associated so that losses occuring earlier in the season count
less than losses which occur later in the season. See Loss Factor
for more details.
Running Up the Score Factor
This factor may be used in the high school CCR to offset gains made in the
Power Ratings by "running up the score". The HS Coaches will determine the
definition of "Running Up the Score" and prescribe a means for measuring this
factor. See Running Up the Score for more details.
Victory Factor
All wins (not just quality wins) count in this component and a time factor
can be associated so that wins that occur later in the season (especially
championship games) count more than earlier season victories. See
Victory for more details.
In summary, the purpose of this new procedure is to correct problems of past
years including not emphasizing winning/losing, placing too much emphasis on
goal margins, running up the score or holding down the score, adding
a time factor for when the game was played and other factors which will be
considered through a coaches poll. The new procedure will allow anyone to
compute exactly, albeit somewhat cumbersome, how their team was rated without
some degree of arbitrariness as was done in years past.
|
Sponsored Links