So, here's how we do it. Follow the links for the detailed computations.
Lots of changes this time. Although TBRW beat the ECAC Coaches Poll for the third time in four years, it was a weak year.
All the stats which used to be expressed in rough thirds (+1, 0, 1) have been changed to avoid being swamped by the other metrics.
RS is no longer the team's place in the standings. It is based on the prior season's winning percentage.
PS is now multiplied by 2 (for number of "points" advanced).
1/2 is renamed Improve, as it confused everybody to have a fraction as a metric name.
Prv3 is gone. It existed to help flatten out the exaggerations caused by the standings metric, which no longer operate.
RetF and RetD are gone, replaced by Off, the returning percentage of points of high performance players, defined below.
RetG is gone, replaced by Goal, the returning percentage of goalie wins.
Recruit is a brand new metric, described below.
Ned replaces both Pred and FPred, as described below.
FPred is no longer necessary since Ned includes recruiting strength. The whole idea of factoring a regression to the traditional standing of a team was to capture its "reload" capability. Now that a targeted recruiting metric is available, this is no longer necessary.
TBRW is now the rank order of Ned rather than the defunct FPred.
Regular season points 

Postseason performance 

Secondhalf improvement 

Returning % of high performance players 

Returning % of goaltender wins 

BRP grade of incoming recruiting class 
Ned (the contraction of "New Predictor") is derived by adding all of these values together:
Ned = RS + PS + Improve + Off + Goal + Recruit
TBRW, the prediction of standing, is the rankorder of Ned, with ties broken as discussed below.
RS = Prior regular season's points, or percentage multiplied by 22 if not all teams played 22 games (a.k.a., the Pachyderm Contingency).

Pct  RS 
Brown 
.364  16 
Clarkson  .682  30 
Colgate  .386  17 
Cornell  .545  24 
Dartmouth  .614  27 
Harvard  .500  22 
Princeton  .500  22 
Quinnipiac  .545  24 
RPI  .386  17 
St. Lawrence  .750  33 
Union  .341  15 
Yale  .386  17 
The number of postseason upset series wins (+1) and losses (1) in the previous season's ECAC playoffs, multipled by 2. The consolation game is ignored.

PS 
Brown 
0 
Clarkson  0 
Colgate  0 
Cornell  2 
Dartmouth  0 
Harvard  0 
Princeton  0 
Quinnipiac  4 
RPI  0 
St. Lawrence  2 
Union  0 
Yale  0 
One third (rounded to a whole) of the difference in points gained in the second half (games 1222) minus the first half (games 111) of the prior ECAC regular season.

Pts Total  Pts 1222  Pts 111  Diff  Improve 
Brown 
16  8  8  0  0 
Clarkson  30  16  14  2  1 
Colgate  17  8  9  1  0 
Cornell  24  11  13  2  1 
Dartmouth  27  17  10  7  2 
Harvard  22  14  8  6  2 
Princeton  22  13  9  4  1 
Quinnipiac  24  9  15  6  2 
RPI  17  10  7  3  1 
St. Lawrence  33  16  17  1  0 
Union  15  9  6  3  1 
Yale  17  6  11  5  2 
Returning % of points of "high performance players," multiplied by 10 and rounded to nearest whole number.
What is a "high performance player," you ask? A player who scored at least half as many points as the team leader in the preceding season. For example, Topher Scott led the 2007 Big Red with 25. Cornell's high performance players are therefore anyone above the Cut, 25/2, thus 13 or more points.
Scott 25, Sawada 21, Greening 19, Gallagher 18, Krantz 15, Seminoff 15, Nash 14 = 127 points are returning (Ret).
Bitz 24, McCutcheon 20, Romano 19, Carefoot 17 = 80 points are not (Not).
Therefore, Cornell's Off% is 127/(127+80) = .614, and their Off is 6.
All prior seniors are assumed to be not returning. In addition, the nonsenior, nonreturning players from last year's ECAC rosters are the following (* high performing players):

Cut  Ret  Not  Off%  Off 
Brown 
17  116  18  .866  9 
Clarkson  20  194  40  .829  8 
Colgate  20  121  59  .672  7 
Cornell  13  127  80  .614  6 
Dartmouth  22  54  148  .267  3 
Harvard  13  131  66  .665  7 
Princeton  15  75  78  .490  5 
Quinnipiac  22  172  41  .808  8 
RPI  14  113  97  .538  5 
St. Lawrence  17  99  89  .527  5 
Union  19  132  58  .695  7 
Yale  16  72  18  .800  8 
Returning % of goaltender minutes multiplied by 4 and rounded to nearest whole number. The relative deemphasis is because teams will often return all or no goaltending, and a ten point swing would weigh far too much in the final rankings.
The nonsenior nonreturning goaltenders from last year's ECAC rosters are:

Ret  Not  W%  Goal 
Brown 
11  0  1.000  4 
Clarkson  24  1  .960  4 
Colgate  15  0  1.000  4 
Cornell  14  0  1.000  4 
Dartmouth  18  0  1.000  4 
Harvard  7  7  .500  2 
Princeton  8  7  .533  2 
Quinnipiac  21  0  1.000  4 
RPI  10  0  1.000  4 
St. Lawrence  23  0  1.000  4 
Union  14  0  1.000  4 
Yale  9  2  .818  3 
Based on the grade assigned by the Big Red Puckhead blog, for ECAC recruiting, (A = 4, B = 3, etc).
Teambyteam breakdown and details are here.

BRPH  Recruit 
Brown 
D  1 
Clarkson  B  3 
Colgate  A  4 
Cornell  A  4 
Dartmouth  A  4 
Harvard  A+  4 
Princeton  B  3 
Quinnipiac  C  2 
RPI  B  3 
St. Lawrence  D  1 
Union  C  2 
Yale  C  2 
Ned = RS + PS + Improve + Off + Goal + Recruit
TBRW
TBRW is the rankorder of Ned. Ties in Ned are broken as follows:
If a tie still remains, then a tie is recorded.

RS  PS  Improve  Off  Goal  Recruit  Ned  TBRW 
Brown 
16  0  0  9  4  1  30  10 
Clarkson  30  0  1  8  4  3  47  1 
Colgate  17  0  0  7  4  4  32  8 
Cornell  24  2  1  6  4  4  35  6 
Dartmouth  27  0  2  3  4  4  40  3 
Harvard  22  0  2  7  2  4  37  5 
Princeton  22  0  1  5  2  3  33  7 
Quinnipiac  24  4  2  8  4  2  40  4 
RPI  17  0  1  5  4  3  30  9 
St. Lawrence  33  2  0  5  4  1  41  2 
Union  15  0  1  7  4  2  29  11 
Yale  17  0  2  8  3  2  28  12 