Never Apologize, Never Explain

So, here's how we do it. Note that we have adjusted Ned a little from last year, dropping the recruiting metric (a noble stat from a great source, but too difficult to weight) and restoring the modifiers from the 2007 Predictions, with the modification of RetG from Wins to Minutes and scaling all percentages.

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

RS Regular season performance in prior season
PS Post-season performance in prior season

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

Imp Second-half improvement in prior season
Ret Returning players from prior season

RS

RS = Prior regular season's points, or percentage multiplied by 44 if not all teams played 22 games (a.k.a., the Pachyderm Contingency).

 RS
Brown 15
Clarkson 33
Colgate 21
Cornell 25
Dartmouth 15
Harvard 27
Princeton 28
Quinnipiac 22
RPI 15
St. Lawrence 16
Union 25
Yale 22

PS

The number of upset advances (+1) or eliminations (-1) in the previous season's ECAC playoffs, multipled by 2. The consolation game is ignored.

  • Colgate upset Clarkson in the QF.
  • Cornell upset Union in the QF.

 PS
Brown 0
Clarkson -2
Colgate 2
Cornell 2
Dartmouth 0
Harvard 0
Princeton 0
Quinnipiac 0
RPI 0
St. Lawrence 0
Union -2
Yale 0

Imp

One-half (rounded down) of the difference in points gained in the second half (games 12-22) minus the first half (games 1-11) of the prior ECAC regular season.

 Pts G 12-22 G 1-11 Diff Imp
Brown 15 10 5 +5 +2
Clarkson 33 18 15 +3 +1
Colgate 21 12 9 +3 +1
Cornell 25 11 14 -3 -1
Dartmouth 15 8 7 +1 0
Harvard 27 15 12 +3 +1
Princeton 28 14 14 0 0
Quinnipiac 22 9 13 -4 -2
RPI 15 7 8 -1 0
St. Lawrence 16 6 10 -4 -2
Union 25 11 14 -3 -1
Yale 22 11 11 0 0

Ret

This metric is based on a returning % of last year's players.

  • F%: returning % of points of top 9 forwards
  • D%: returning % of points of top 4 defensemen
  • G%: returning % of minutes of all goalies

We sum these three stats and then multiply by 7 to scale the percentages to a metric comparable to the other season points metrics. We chose 7 by the following arbitrary logic. The difference between the theoretical max and min of the sum of the three stats is 3.00. Multiplying by 7 gives a theoretical maximum spread of 21 points -- just barely shy of one-half of an ECAC season. That seemed about right. Note that in practice the real spread will be around 1.00, or 7 points -- very close to the max spread of 6 points from the former RetF, RetD and RetG stats.

  • Ret: 7*(F% + D% + G%)

The team breakdowns are shown here.

All of last year's seniors are assumed non-returning. In addition, the follwing players who would qualify by these categories are known to be non-returning:

  • Kyle Richter, G, Harvard

 F% D% G% Ret

Brown

 .66 .10 1.00 12
Clarkson .62 .50 .07 8
Colgate .60 1.00 .00 11
Cornell .69 .85 1.00 18
Dartmouth .44 1.00 .15 11
Harvard .48 .66 .02 8
Princeton .93 .63 1.00 18
Quinnipiac .62 .88 .79 16
RPI .70 1.00 .61 16
St. Lawrence .79 1.00 .45 16
Union .81 .86 .56 16
Yale .75 .85 1.00 18

Putting it All Together

Ned

Ned = RS + PS + Imp + Ret

The predicted finish is the rank-order of Ned. Tie-breakers go in order of the stats: RS, PS, etc.

 RS PS Imp Ret Ned Rank
Brown 15 0 +2 12 29 11
Clarkson 33 -2 +1 8 40 3
Colgate 21 +2 +1 11 35 8
Cornell 25 +2 -1 18 44 2
Dartmouth 15 0 0 11 26 12
Harvard 27 0 +1 8 36 6
Princeton 28 0 0 18 46 1
Quinnipiac 22 0 -2 16 36 7
RPI 15 0 0 16 31 9
St. Lawrence 16 0 -2 16 30 10
Union 25 -2 -1 16 38 5
Yale 22 0 0 18 40 4