I like the idea of an "expert panel" deciding on what a player's defensive rating should be.
That may be an idealogically objectional opinion to those who believe that statistics should be the only data used in the sim engine.
I am not sure how to measure a player's defensive abilities with statistics, though.
Some ideas:
1) plus/minus,
2) minutes spent on the PK as a percentage of a team's total PK time;
3) minutes of ice time during the last 10 minutes of the third period during times the team has a 1-goal lead, as percentage of the total of all such time for the team;
4) overall team defensive record;
5) body checks per minute of playing time;
6) blocked shots per minute of playing time;
7) calculating the goals-per-minute scored by all 5 opposing skaters when playing against Player X vs. the goals per minute those opposing players scored against the rest of the league that season;
8) # of GA on the ice for per minute of playing time.
Problems with above ideas:
1) +/- also measures offense...Say a forward line combines for 140 goals, and is on the ice for 105 goals against, then they are +35. * Another forward line, with the same amount of ice time, combines for 65 goals and is on the ice for 50 GA, then they would be +15. But it could be argued that the second line, due to the GA data, is the better defensive line. (* This example assumes only regular-strength shifts.)
2) I like this one. But subjective to the coaches preferences. And PK is but one aspect of defensive play.
3) Love this one.
4) A common criticism with +/- goes something like "Billy-Bob, Frank, Isiah, and Ruprecht were playing perfect positional hockey, but Stanky missed his check and so they all get a minus."
Using the Team Defensive Record has the exact same issue, except that when Stanky misses his check everybody on the bench "gets a minus", (or whatever), too.
5) Like this one, but again, body checks measure only one (of several possible) defensive styles.
6) Narrow. A piece of the puzzle?
7) Is this even possible to do?
8) This stat does not measure how players are chosen by the coach...e.g., if you are your team's best defender, you may be out all the time against Peter Statsny, Dale Hawerchuk, Wayne Gretzky, Mike Bossy, and Denis Savard. One would expect such a player to be on the ice for more GA.
So, as many questions as possible answers.
In general, I really like the hockey product produced by WIS. I do think a measure of a player's defensive acumen is important, though, for accurate simulating. I also think this defensive measure should measure only defensive aspects of the game (not, for example, PIM, which can be simmed for independently).