The pivotal question in this post is whether the highly subjective feeling of Cornell "being hot" or "being on a roll" in the second half of the season is a product of mere perception or statistically identifiable changes between performances in the first half of the season compared to the second half of the season. I applied a similar statistical regime that I outlined in Part One of "The Better Half" of this series yesterday to the entire landscape of college hockey programs over the last two seasons. The focus on the performance of a specific team or program allows for greater depth of analysis.
This post focuses upon the Schafer Era at Cornell. A punctuated change in the complexion and composure of Cornell hockey occurred when Mike Schafer became head coach of Cornell for the 1995-96 season. This change is why this post focuses upon the last 17 completed seasons that Schafer has been Cornell's bench boss.
The bellwether events of the Schafer Era that lend themselves to the cursory conclusion that Cornell is a program geared for the second half and post-season of a given season are the winning of five Whitelaw Cups, earning of one berth to the Frozen Four, and garnering of nine berths to the NCAA Tournament over 17 seasons. Cornell has advanced to the ECAC Championships as one of the final four teams in the ECAC Tournament in each of the last five ECAC Tournaments. That feat includes Schafer's and Cornell's turning around of a 2010-11 season that was heavily rebuilding with tallying only four wins in the first half of the season, but making a run to the 2011 ECAC Championship Final game.
Statistics yield a different conclusion when comparing early season results to overall season results. This was the approach taken in Part One of "The Better Half." The halfway point in a season again was defined as December 15. This choice ensures that Cornell's results in the Florida College Hockey Classic are tabulated as only part of second-half winning percentages or overall season winning percentages. The results of analysis of all 17 completed seasons of the Schafer Era below are shown and discussed.
Closer analysis undermines this point. Several of the data points that appear above the line in the graph are more appropriately on the line. The final count of data points above the line and below the line would yield seven and ten data points respectively. This requires a more nuanced approach to determine the accuracy of the belief that Cornell is a program that improves its second-half performance regularly.
The various data points in the graph indicate the accolades that Cornell earned in a given season. The five red, circular data points represent Cornell's five ECAC Championships during the Schafer Era. The four data points depicted with white diamonds outlined in red represent Cornell's four at-large berths to the NCAA Tournament during the Schafer Era. The single Frozen-Four berth during the Schafer Era occurred in 2003.
The graphical depiction above confirms readily that ten of Schafer's 17 seasons witnessed Cornell's earning of an overall season winning percentage at the close of competition below the winning percentage that Cornell owned at the midpoint of the associated season. The graph confirms however that when the Big Red wins an ECAC Championship that it typically ends the season with a winning percentage that is greater than the winning percentage it enjoyed at mid-season point. Cornell had a better overall winning percentage in its ECAC Championship-winning years of 1996, 1997, and 2005 than its mid-season winning percentage. The ECAC-Championship years of 2003 and 2010 saw Cornell underperform in its overall season winning percentage relative to its winning percentage at the midpoint. All four of the years that Cornell earned an at-large berth to the NCAA Tournament, Cornell did not match the pace of its early season winning percentage in its ultimate, overall season winning percentage.
The fact that the weightiest part of Cornell's schedule falls in the second half of each season in terms of number of games played indicates that overall season winning percentage is highly correlated with second-half winning percentage. Despite the likelihood of the data being nearly identical to that shown above for overall season, I provide the comparison between early season winning percentage and second-half winning percentage in the above graph.
The graph predictably mirrors in trend the graph above it comparing the difference between overall season and early season winning percentages. Three of Cornell's five ECAC Championships during the Schafer Era came when Cornell improved its second-half winning percentage relative to its early season winning percentage. The remaining two ECAC Championships that Cornell has won during the Schafer Era came in 2003 and 2010 when Cornell did worse in terms of winning percentage in the second half of the season. Cornell's four at-large berths to the NCAA Tournament over the last 17 seasons came in seasons when Cornell underperformed in the second half of the season.
The decline from an early season winning percentage of 0.909 in 2003 is all but expected. The 2002-03 team earned a second-half winning percentage of 0.820. It would have been nearly impossible to keep pace with an above-90% winning percent as the longest part of a season beings. The 2009-10 season and 2010 ECAC Championship team is somewhat more puzzling. It began the season with the second-best early season winning percentage of a Cornell ECAC Champion team with a winning percentage of 0.727. That figure is tied with that of 2005 when in the 2004-05 Cornell outperformed its early season winning percentage as shown in the above graph.
The distinction between the 2005 and 2010 ECAC Championship teams at Cornell is likely a product of difficulty of schedule. The 2004-05 season saw Cornell meet only Boston College, Maine, Ohio State, and Minnesota in the second half of the season. It is important to remember that in 2005 Jerry York at Boston College had not begun his current streak of utter dominance in college hockey and his near unbroken chain of Frozen-Four appearances. The second half of the 2009-10 season observed Cornell confronting Colorado College, New Hampshire, and North Dakota. This to some is an unsatisfactory explanation but it can explain how a team whose early season winning percentage was high but not astronomical could underperform in the second half of the season, but appear so dominant that it earned the 2009-10 team the appellation of "the dream-crushing, soul-devouring juggernaut."
The final analytical lens to apply to the collected data is to determine trends beyond comparison of mere data points. When one includes all 17 completed seasons of the Schafer Era, Cornell's average absolute change in winning percentage between the early season and the overall season winning percentage is a reduction of 0.027. This all-inclusive average absolute change in winning percentage of -0.027 is tempered by a standard deviation of 0.119. The typical range of variation for the difference between Cornell's mid-season winning and its ultimate, overall winning percentage is from -0.146 to 0.092.
There is a statistical difference between the range of winning percentage differential between the early season and overall season if one analyzes Cornell's championship seasons in isolation. The average change in winning percentage from early season to the overall season results is an improvement of 0.024 when controlled for ECAC Championship seasons. The associated standard deviation is 0.083. The standard deviation for championship-winning seasons is appreciably smaller than the standard deviation associated with all seasons during the Schafer Era. This produces a range of winning percentage differential between the first half and the overall season-ending winning percentage of from -0.059 to 0.107.
The typical reason for performing extensive statistical tests or analysis is to apply them to a predictive model. However, one thing that the above trends indicate is that the ultimate tone of a season cannot be reduced simply to winning percentages or improvements upon early season results. The 2010-11 season was successful in its own right with a rebuilding team that made an unexpected run to the 2011 ECAC Championship Final, however, it will not be remembered like the 2002-03 season or even the 2009-10 season in the distant future even though the 2010-11 season witnessed one of Cornell's greatest improvements over early season winning percentage.
Nonetheless, application of the above typical ranges of the differential between early season and overall season winning percentages to the 2012-13 season will provide a metric by which some members of the Faithful can measure the success of this Cornell team. Cornell completed the first half of the 2012-13 season with a winning percentage of 0.636. If one desires to assume that this season is a normal season in the Schafer Era, that would lead one to conclude that the season's overall winning percentage would end in the range of 0.490 to 0.728. If one embraces the opinion of WAFT that this season is not a typical season and hopes that this may be a championship year, then an overall season winning percentage in the range of 0.577 to 0.743.
Cornell is a great program and has been a great program in the Schafer Era not because of its ability to improve its performance in the second half of the season unequivocally. The fact that ten seasons of the Schafer Era saw decreases in winning percentage in the second half prove that. The reason Cornell is a great program and returned to greatness under Schafer is because of the program's and his ability to get results when they are needed most. Cornell as a program has an uncanny ability to dose its efforts in a manner to win the games that it needs to win. That is why it is a dominant team in the playoffs, not because it holistically performs better after the semester break.
It takes four wins to win an ECAC Championship if a program earns a first-round bye. It takes four wins to win a national championship. The key to success of the 2012-13 season will be found not in statistics but in the team's ability to rise to the challenge of its opponents and see in adversity opportunity. This team has the ability to win those eight games for ultimate glory. However, this team needs to continue to prove that it can get key wins when they are needed. This weekend's series against Denver is key.