One of the things that makes computerized rankings compelling is that teams are ranked on well defined criteria rather
than the conventional wisdom/groupthink that is often exhibited in ranking polls. Such rankings give another viewpoint
to consider.
This ranking algorithm has proven itself over time against some of the most informed and specialized hockey "polls"
around. These rankings have done consistently better than the coaches' seedings (at the end of the regular season the
head coaches of all teams in each section meet to seed the (8-15) teams in their section for the playoffs) at picking playoff winners
in Minnesota and Wisconsin high school hockey (despite home ice being awarded based on the seedings). So the rankings should
at least raise the question of whether they could be right whenever they differ from accepted thought.
Having said all of that, I doubt that many people would rank Shattuck over Team Illinois based on YTD results, but
the rankings program still does based on its objective criteria. We can think the rankings are wrong in this instance,
and they may be, but are we really sure which of the two teams is better (think "which team would win a 10 game series between
the two"?). I have no idea what the answer is (and that may be a bad analogy as there is no accounting for matchups in the
rankings and the rankings are based on the full season of scores against all teams), but the computer rankings are rating Shattuck
higher at the present time. It will be interesting to see how these two teams do against each other throughout the rest of the
year (they will probably meet another half a dozen times) to see if the rankings are 'prescient' or 'blind' in this case.
For those interested, I offer the following analysis to help us understand what goes into the rankings, how this past
weekend's games affected them, and point out what the weak link in the rankings is when comparing these two teams:
SSM came into the weekend rated .2 goals ahead of TI. When two teams have a strong schedule interlock (many common
opponents, etc), it is fairly easy to estimate what effect a few games will have on their relative rankings by calculating
an approximate per game performance (goal diff + opponent rating) for the weekend and then using the ratio of new games to
total games to find out how much to add or subtract from their ranking. TI's average game performance for the weekend comes
out to 13.19 and SSM's to 11.66. TI's season ranking would then be approximated to go up by (13.19 - previous rating(=11.63))
*3/25=~.19, and with SSM approximated to go down by (11.66 - previous rating(=11.83))*3/18=~.03). So, with TI moving up by
.19 and SSM down by .03 TI should then be rated .02 ahead of SSM, right?
Not so fast. Unfortunately, it is a bit more complicated than that as SSM and TI had a very weak schedule interlock
coming into the weekend. Three games each by the Pittsburgh Hornets and Ojibwa Eagles against teams from the Minnesota
league were all that connected teams in that league to the teams outside of that league. With so few games connecting the
two scoresets, the other games of the weekend could easily move the rankings of the teams in that league relative to those
outside it.
The other team from that league that played this past weekend was Team Wisconsin. The two new teams for the weekend,
Minnesota Red and Minnesota White, serve as a common opponent of teams in the league and teams outside it and do help link
the league to those teams outside it.
Team Wisconsin's outcome against TI was almost a goal (.85) better than predicted by last week's ratings. MN White & Red
played three games against teams from the league and three games against non-league teams. Their average game performance
against teams from the league was 1.62 goals less (10.62 vs. 12.24) than their average performance against teams from
outside the league. These seven games will obviously result in the Minnesota league as a whole being elevated somewhat
from the week before as compared to teams outside the league as they will outweigh the fact that SSM underperformed based
on the previous week's ratings by an average of 1.06 goals in their two games against non-league opponents.
It would be extremely difficult to calculate by hand the exact amount that each of the tightly schedule-bundled teams in
the league would be elevated from their previous ranking based on the huge web of scores that go into the rankings but,
after comparing the two weeks rankings, it would appear that the amount turns out to be ~.21 goals as that would explain TI
still being .19 behind SSM.
To summarize what happened (without all of the numbers), it appears that a strong performance by the regular season last
place team in SSM's league (Team Wisconsin) and two common opponents (Minnesota Red and Minnesota White) that had much
better game performances (based on opponent rankings and game scores) against teams from outside of that league than against
those in it, actually moved the Upper Midwest Elite League Teams (tightly schedule interlocked with each other) up even more
than Shattuck's lower than expected performance moved them down. Along with keeping Shattuck ahead of team Illinois, this
upward movement of the teams in that league actually moved idle teams in the league up ahead of other teams that they had
been very close to in the previous week's rankings.
The insular schedule of the Minnesota leagues makes their rankings have a greater margin of error than the rankings of
teams outside of those leagues. As a result, their rankings are much more likely to change based on a few new non-league
results. This was seen with the Elite 2 league two weeks ago when a poor showing by four of those teams against midget
teams caused teams in that league to fall 40 or more places in the rankings in one week.
There was a note about the weak schedule interlock on the rankings earlier in the season and a new note appears on this
week's ratings. In any case, most of the teams from that league are done for the season and will be removed from the
current rankings later this month. SSM Prep is the only team from that league that will play between now and February.
Team Wisconsin, the only other team from the league that will play again this season, will have their ranking rise and fall
until they resume play largely based on the fortune of SSM Prep's games against non-league teams during the rest of the
season.
For a more detailed description of how the rankings actually work, follow the Rankings Key
and About the Rankings links.
©2005 MidgetAAAHockey.com. All rights reserved.