The first graph included here represents a comparative analysis between each Montreal Canadiens player's risk/reward rating through the first 24 games of the season and their risk/reward rating over the course of the season's final 24 games.
The second graph compares the risk/reward rating produced through the fist 24 games of the season to their end-of-year risk/reward rating.
A correlation between the two variables in each graph will demonstrate two things. Firstly, it will tell us whether 24 games of data is enough to reasonably predict a player's future performance. Secondly, it will demonstrate how much consistency there is in my tracking methods.
COMPARING FIRST HALF R/R TO SECOND HALF R/R
The higher the r-squared value, the higher the relationship between the 2 variables. The first-24 to second-24 comparative analysis indicates an r-squared value of 0.8197. In theory, this tells us that 82% of a player's second-half performance was predictable through 24 games.
Lars Eller produced the biggest difference between his risk/reward after 24 games, and his risk/reward over the final 24 games. Eller went from a risk/reward rating of 1.60 over the first 24 games, to an incredible 2.02 over the final 24 games. The bulk of Eller's improvement came in the final 10-games of the season; 10 games that saw him score 4 goals while adding 7 assists.
When we remove the two biggest outliers within the collected data, the r-squared value jumps to 0.870.
Only players with at least 150 even-strength minutes were included
COMPARING FIRST HALF R/R TO END-OF-SEASON R/R
A comparative-analysis of each player's risk/reward earned through 24 games to their risk/reward rating at the end of the season produced an r-squared value of 0.9453. This tells us that 95% of a player's final risk/reward rating was predictable through 24 games. If we remove the two outliers within the data the r-squared value jumps to 0.973.
