Comparing year-to-year risk/reward ratings provides us not only with the opportunity to compare each player's performance last year with the current season, it also provides us with an opportunity to test the accuracy and repeatability of data-generated scouting. It also allows us to evaluate possible future increases or decreases in players' ratings; as determined by their past performance.
This simple bar graph represents the risk/reward rating each Montreal Canadiens player produced last season, as well as their rating from the current season. The graph matches up well with how most people would view the current players, as it shows (in its simplest form) that Josh Gorges and Andrei Markov have played better this year, while David Desharnais, and Max Pacioretty have struggled through most of this season.
That said, from my perspective the more important result communicated in this graph is the similarities from season to season.
Comparative-analysis
The purpose of the graph below is to compare the relationship between the risk/reward ratings produced in 2012-13 with those produced this season. Only those players with at least 150 even-strength minutes of ice-time have been included in this analysis. The reason for this is that it usually takes 150 minutes of play to stabilize a player's rating.
At its simplest form, this graph is meant to communicate accuracy and repeatability. The higher the r-squared value the closer the relationship between the 2012-13 risk/reward ratings and the 2013-14 ratings. A direct relationship would produce an r-squared value of 1. The lower the rating from this point, the less pronounced the relationship.
This r-squared value of 0.872 tells us that 87% of a player's risk/reward is related to the risk/reward they produced the year before.
If we use this to project the likely future performance of a Rene Bourque's future risk/reward will likely fall some where between 0.94 and 0.72. Josh Gorges risk/reward rating is likely to regress toward 2.10 in the future, while Max Pacioretty's rating is likely to improve toward 1.50.
This process has provided me with results that reinforce my faith and pride in the system I've developed. Through the first quarter of the 2013-14 season I have tracked just under 30,000 points of data. Last season I tracked almost 60,000 points of data. Incredibly enough, through all of that data, the overall team risk/reward ratings produced by combining every Montreal player last year was 1.67; exactly the same as this season.