It should be noted that in other sports, including the NHL, the correlation between playing well at the end of the season and doing well in the playoffs is weaker than the correlation between the record over the entire season and doing well in the playoffs. This becomes especially true if you control for late changes to teams' rosters, such as pickups at the trade deadline and those who return from or are lost to injury. For sports that have some sort of projection system based upon roster (such as PECOTA or ZIPS for major league baseball; VUKOTA for NHL; and so on) the best predictor is is using the entire season's worth of data to make a projection based off of the current roster. Being hot at the end of the season is not very useful as a metric going forward.
This is really just a subset of the fact that streaks, of almost any variety, make lousy projections. Winning streaks, or even hot periods, are obvious in retrospect but they do little to tell you who is going to win the next game. The same is true with hot streaks. Whether it's a team's record over the last ten games or a hitter's performance in his last 30 plate appearances, the prediction value is close to nil. The problems of small sample size overwhelm any real element of momentum that might be present.
As a general rule, we should be wary of looking at things in one sport and assuming that it's true in a different sport that's never really been studied. However, the above is true in every sport I am aware of where the question has been asked so my default assumption is that it's probably true in NCAA Division I women's ice hockey as well. I'm not going to overturn that default assumption based upon the observation that on the three occasions that the WCHA regular season and playoffs champions were different teams, it was the playoff champion that won.
This becomes even more true when you actually look at the three cases. In 2000-01, the regular season champion (Minnesota) did not get an invitation to the NCAA tournament, so the hypothesis that winning the conference tournament is a more valuable predictor wasn't even tested. So we really only have a sample size of two, not three.
In 2008-09, Minnesota finished one point ahead of Wisconsin in the regular season standings. What that means is that once the WCHA tournament had been played, Wisconsin actually had a better record in conference games than Minnesota did. Since the claim isn't that regular season conference games make a better predictor than playoff results (I suspect that they do, but since the ratings include all games that really has nothing to do with the claim being made), but rather that using all games prior to the NCAA tournament make a better predictor than the much smaller set of tournament games, there actually isn't a conflict here. The team with the better total season won the championship.
2011-12 is slightly different, but not by much. Wisconsin won the regular season title by two games, 23-3-2 to 21-5-2. After the conference tournament, the records were: Wisconsin 25-4-2; Minnesota 25-5-2. So Wisconsin still had a better record against WCHA opponents overall but the difference was only one game. At that point I would bet that the ratings systems would have suggested that a championship game between the two teams (which is what happened) was pretty much a tossup.
So when the examples are analyzed, pokechecker hypothesis hangs on just one instance in which a game between two almost identical opponents was won by the team that had lost one more game than the other. That's . . . not exactly convincing.
I'm not sure what you read that you condensed down to that sentence, but that isn't how I understand Rutter Rankings at all.
) Of course humans on the selection committees for basketball and pollsters in football certainly are influenced by information from past seasons, far more than the influence of the prior in the Rutter rankings. If the teams who are #1 and #2 in preseason football go undefeated, pretty much no one else has a shot at the championship in the current system.