Never Apologize, Never Explain, 2005

So, here's how we do it.  Follow the links for detailed computations.

Theory...

First we take base values directly computed from last year's team performance:

RS

 Regular season performance in previous season.

PS

 Post-season performance in previous season.

 

Factored into these base values is a function of the following modifiers:

Prv3

Standing in total conference record in the three previous seasons.

1/2

Second-half improvement, previous regular season.

RetF

Returning % of Leading Forwards points.

RetD

Returning % of Leading Defensemen points.

RetG

Returning % of Leading Goaltenders wins.

 

Pred is derived by adding all the base values and modifiers together:

Pred = (RS + PS) + (Prv3 + 1/2 + RetF + RetD + RetG)

FPred, the basis for our final predictor, is derived by regressing Pred 25% back towards their conference standing of the previous ten seasons:

FPred = 3 * Pred + Prv10

Pred

 Sum of all base values and modifiers.

Prv10

 Standing in total conference record in the ten previous seasons.

FPred

 Final predictor and basis for the TBRW? Prediction.

Our final prediction, TBRW, is the rank order of FPred.

TBRW

 Predicted ranking.

The modifiers are computed for the present year, below.


... and Practice

RS

RS is a reversal (returning the largest number for first place) of the regular season Standing of 2003-04.  RS = (12 - Standing).

 

 

Standing RS

Brown

3 9
Clarkson 9 3
Colgate 1 11
Cornell 2 10
Dartmouth 4 8
Harvard 6 6
Princeton 12 0
RPI 5 7
St. Lawrence 10 2
Union 8 4
Vermont 11 1
Yale 7 5

 

PS

The number of post-season upset series wins (+1) and losses (-1) in the 2003-04 playoffs.  The consolation game is not counted.

2003-04  upsets:

 

 

PS

Brown

-1
Clarkson +3
Colgate -1
Cornell -1
Dartmouth -1
Harvard +2
Princeton 0
RPI 0
St. Lawrence +1
Union -1
Vermont 0
Yale -1

 

Prv3

The third (+1 top 33%, 0 middle 33%, -1 bottom 33%) in conference winning percentage (Pct) for the past three seasons (2001-02 through 2003-04): the seasons in which the current returning classes have played.

 

Pct Rank Prv3

Brown

.561 4 +1
Clarkson .500 6 0
Colgate .553 5 0
Cornell .788 1 +1
Dartmouth .576 3 +1
Harvard .606 2 +1
Princeton .303 11 -1
RPI .462 8 0
St. Lawrence .409 10 -1
Union .455 9 -1
Vermont .288 12 -1
Yale .500 0 0

 

1/2

The third (+1 top 33%, 0 middle 33%, -1 bottom 33%) in improvement in points gained in the second half of the RS (games 12-22) over the first half (games 1-11), last season (2003-04).

 

Pts Total Pts 12-22 Pts 1-11 Pts Diff 1/2

Brown

28 11 17 -6 -1
Clarkson 18 6 12 -6 -1
Colgate 30 16 14 +2 +1
Cornell 29 14 15 -1 0
Dartmouth 27 13 14 -1 0
Harvard 22 11 11 0 0
Princeton 12 2 10 -8 -1
RPI 27 14 13 +1 0
St. Lawrence 17 10 7 +3 +1
Union 19 13 6 +7 +1
Vermont 15 13 2 +11 +1
Yale 20 8 12 -4 -1

 

RetF

The third (+1 top 33%, 0 middle 33%, -1 bottom 33%)  in returning % of Leading Forwards points.  Team-by-team breakdown here.

 

RetF% RetF

Brown

71 -1
Clarkson 81 0
Colgate 86 +1
Cornell 81 0
Dartmouth 95 +1
Harvard 76 0
Princeton 91 +1
RPI 74 0
St. Lawrence 72 -1
Union 96 +1
Vermont 72 -1
Yale 70 -1

 

RetD

The third (+1 top 33%, 0 middle 33%, -1 bottom 33%)  in returning % of Leading Defensemen points.  Team-by-team breakdown here.

 

RetD% RetD

Brown

25 -1
Clarkson 69 +1
Colgate 64 0
Cornell 77 +1
Dartmouth 81 +1
Harvard 50 0
Princeton 36 -1
RPI 65 +1
St. Lawrence 41 -1
Union 57 0
Vermont 54 0
Yale 23 -1

 

RetG

The third (+1 top 33%, 0 middle 33%, -1 bottom 33%)  in returning % of Leading Goaltenders wins.  Team-by-team breakdown here.

 

RetG% RetG

Brown

0 -1
Clarkson 94 0
Colgate 82 0
Cornell 100 +1
Dartmouth 100 +1
Harvard 100 +1
Princeton 100 +1
RPI 5 -1
St. Lawrence 100 +1
Union 100 +1
Vermont 100 +1
Yale 100 +1

 


Putting it All Together

Pred

Pred = (RS + PS) + (Prv3 + 1/2 + RetF + RetD + RetG)

 

RS PS Prv3 1/2 RetF RetD RetG Pred

Brown

9 -1 +1 -1 -1 -1 -1 5
Clarkson 3 +3 0 -1 0 +1 0 6
Colgate 11 -1 0 +1 +1 0 0 12
Cornell 10 -1 +1 0 0 +1 +1 12
Dartmouth 8 -1 +1 0 +1 +1 +1 11
Harvard 6 +2 +1 0 0 0 +1 10
Princeton 0 0 -1 -1 +1 -1 +1 -1
RPI 7 0 0 0 0 +1 -1 7
St. Lawrence 2 +1 -1 +1 -1 -1 +1 2
Union 4 -1 -1 +1 +1 0 +1 5
Vermont 1 0 -1 +1 -1 0 +1 1
Yale 5 -1 0 -1 -1 -1 +1 2

 

Prv10

Prv10 is a reversal (returning the largest number for first place) of the regular season Standing of the previous ten seasons, 1994-95 through  2003-04.  Prv10 = (12 - Standing).

 

Pct Standing Prv10

Brown

.422 11 1
Clarkson .661 1 11
Colgate .562 3 9
Cornell .624 2 10
Dartmouth .438 9 3
Harvard .523 5 7
Princeton .429 10 2
RPI .521 6 6
St. Lawrence .539 4 8
Union .361 12 0
Vermont .459 8 4
Yale .461 7 5

 

FPred = (3 * Pred + Prv10).

The standings prediction, TBRW, is the rank order of FPred.

 

Pred 3*Pred Prv10 FPred TBRW

Brown

5 15 1 16 7
Clarkson 6 18 11 29 5
Colgate 12 36 9 45 2
Cornell 12 36 10 46 1
Dartmouth 11 33 3 36 4
Harvard 10 30 7 37 3
Princeton -1 -3 2 -1 12
RPI 7 21 6 27 6
St. Lawrence 2 6 8 14 9
Union 5 15 0 15 8
Vermont 1 3 4 7 11
Yale 2 6 5 11 10