As a refresher for the CCR calculations, a simplified example is helpful for ease of disposition. Let's assume that 1) NU plays a game on a neutral court against iLOLinois, 2) that after the game results are incorporated that kenpom reports the iLOLini have an adjusted offensive efficiency and an adjusted defensive efficiency that are both equal to 1.0 within B1G play, and 3) NU wins 110-105 in a 100 possession 3OT thriller.
NU would have a CCR-O equal to 1.1 (NU's PPP) less 1.0 (iLOLini AdjD) or 0.1
NU would have a CCR-D equal to 1.0 (iLOLini AdjO) less 1.05 (iLOLini PPP) or -0.05
NU would have a CCR equal to the sum of 0.1 CCR-O and -0.05 CCR-D or 0.05
In effect the CCR takes into consideration not only how a team has performed on an efficiency basis but also the strength of the competition. Note that the CCR calculations also take into consideration home court advantage but we've left that part of the calculation out of the above hypothetical for simplicity.
Before we used a trend line analysis for each team's CCR as the basis for power ranking. Upon further reflection we feel that the use of a trend line analysis -- while informative and provides useful insight to how teams are improving or regressing as the B1G season progresses -- is inferior to the use of an averaging approach that is less susceptible to statistical significance concerns not to mention a greater stability in the rankings and less dependence on early B1G conference games results. Our thinking on the averaging approach is still evolving. Right now we are using a simple average but are also considering an approach that places more weight on recent games.
Below is a chart that shows the current B1G power rankings based on average CCR. We also show in the chart the old trend line analysis as well as the best and worst CCR performances by each team. We've recently tweeted about some of the outlier games, but see notes for the most up-to-date information.
