TBRW Predictions: 2001

2000 was a breakout year for the ECAC, with St. Lawrence reaching the Frozen Four and Colgate only denied a trip to the QF and beyond by a blown call.

2001, on the other hand, could be the breakout year for Cornell.  While the other contenders have dropped back a bit due to massive graduation hits, the pillar of the Big Red (an outstanding and intimidating defense) returns intact.

We see a dog fight for the top slot, with Clarkson returning to form and just nosing out Cornell.  St. Lawrence and Colgate slip a bit, but still look threatening.

 

 

Team

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

 


How We Do It

The full statistical analysis of our predictions is included under "Never Apologize, Never Explain," below.

We begin by computing a number of points for the coming season based on the teams' recent performance and the players they are returning.

In many cases over the past seasons, this base stat, Pred, has out-performed the ECAC Coaches' Poll.

There are times, however, when Pred has failed to predict a big move in the standings for one team or another.  In most cases, this has been a failure to predict a traditionally strong team returning to form after a few lean years. In other cases it has been a traditional also-ran, falling dramatically after one or two seasons in the sun, which has foiled our predictor.

Our enhanced predictor, HPred takes into account the past history of the ECAC members, as well as the statistical effect regression to the mean. 

HPred (Historical-Predictor) factors in the historical strength of each ECAC team over the history of modern Div. I ECAC play (by league records, since the 1964-65 season). Effectively, the Pred stat is pulled or pushed half the distance back towards the lifetime winning percentage of the program (there are some normalization effects which we need not go into here).

It is the final results of HPred that dictate our predicted standings, above.

The full results of HPred are as follows:

 

Team

W L T Pts
1 Clarkson 16 6 0 32
2 Cornell 14 6 2 30
3 St. Lawrence 14 7 1 29
4 Colgate 11 8 3 25
5 Vermont 10 9 3 23
6 Harvard 10 10 2 22
7 RPI 10 11 1 21
8 Princeton 8 10 4 20
9 Dartmouth 7 12 3 17
10 Brown 7 13 2 16
11 Yale 6 12 4 16
12 Union 6 15 1 13

Note that in order to equalize the records at a mean of 22 pts, one point was added for the school whose long HPred required the least upward adjust to reach .5.  This turned out to be Cornell, whose 29.39 required just a 0.11 adjustment.


Assumptions

These metrics are based on certain players leaving early or coming back.

Note that the following non-seniors are known to be leaving early:

These non-seniors who left during last year, and thus have partial year statistics, will not return:

Finally taken into consideration are players active in 1998-99 who missed 1999-2000 but will return for 2000-01:


"Never Apologize, Never Explain"

Base Values

The TBRW? Predictions are based on each team�s previous year�s performance, modified by off-season roster changes.

Each team starts with its previous regular season standing RS.

The number of post-season upset wins or losses PS is then added.

RS  Regular season pts (previous year).
PS  Post-season performance (previous year).

Modifiers

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

5pct  Conference winning pct in 5 seasons previous to past season
1/2  Second-half improvement, previous regular season
RetG  returning % of previous season's goals scored
RetD  returning % of previous season's defenseman games played
RetM  returning % of previous season's goaltender minutes played

Computed Metrics

The coefficient multiplier for the sum of the modifiers helps balance the effects of roster changes and also generates Pred, the former metric which signified the number of points a team should finish with.

Finally, the newest cast member of our statistical stage, HPred, is calculated. HPred is easily derived from Pred when you have the career records of each team in conference play, EPct.

The average of EPct is slightly different from .500, owing to changing in the length of the schedule and conference membership over its history. Therefore, we also compute MEPct, a modification of EPct that pulls the average EPct back to .500. MEPts, meanwhile, represents the number of points you would get with the team's traditional conference percentage, normalized for the relative strength of current conference members, over a full 22 game slate:

MEPct = EPct*(.500 / Mean[EPct]) MEPts = MEPct * 44
  Finally, HPred is then derived from simply averaging the old stat, Pred, and the historical measure, MEPts. HPred = (Pred + MEPts)/2

Pred  Summary metric of previous measures.
EPct  ECAC Historical Pct.
HPred  Putting it all together!