Saturday, July 20, 2013

Impact of Quality of Competition on even-strength risk/reward rating

No one system can place all-encompassing value on a player's performance. But, by combining multiple systems, and including all possible tangible aspects of an individual player's play, we can push advanced stats in the right direction; a direction that provides insight as well as context.

In a constant effort to improve my player tracking system, I've looked into more traditional advanced-stats metrics in order to see if there was anything that could by applied to my system.

Fenwick and Corsi numbers take a global picture of what occurs when a player is on the ice (shots attempted for and against), and uses those numbers to rate individual players. My system tracks individual player's puck-possession successes and failures (passes, dekes, puck-battles, etc) with a similar goal in mind.


 Habs Eye on the Prize's Andrew Berkshire explains Corsi as follows:

Corsi - is a +/- statistic for a player/team that measures all shot attempts, including misses and blocked shots, directed for and against the team/player being measured per 60 minutes.

Andrew explains Corsi Relative Quality of Competition as follows:
 
Relative Corsi quality of competition - a measure of the average relative Corsi score of the opponents a player faces, weighted against the ice time played against each player


Silversevens.com explains the calculation for Corsi Relative Quality of Competition as follows:


Corsi Rel QoC is the weighted Relative Corsi Number of a player's opponents.
For example, if a player plays 30% of the time against five players with a relative corsi of +1.5, 35% of the time against five players with a relative corsi number of +0.2, and 35% of the time against five players with a relative corsi number of -2.1 then:

Corsi Rel QoC = (0.3 * 5 * 1.5) + (0.35 * 5 * 0.2) + (0.35 * (5 * (-2.1)) = -1.075



The top graph is a visual representation of each Montreal Canadiens player's even-strength risk/reward rating as calculated using my system, while the other bar is a visual representation of each player's ES risk/reward rating after including Corsi relative quality of competition into the calculation.

Even-strength risk/reward is a rating that determines how many more positive events than negative events a player produces per-minute of even-strength ice-time. Events used within this system include; puck-battles, loose-puck recoveries, passes, dekes, shots, blocked passes, blocked shots, dump-ins, dump-outs, and deflections. Each event is tracked by success or failure.

In order for the calculation to work, and for the numbers to make sense, I've divided each Canadiens player's Corsi Rel QoC number by 5. Not only does this help minimize the impact on the original number, it also relates better to the reality of my system; Corsi uses team numbers while a player is on the ice. Because of this it multiplies the value by 5 to represent the five skaters on the ice. My system tracks individual events taking place against individual players. As such, it does not need to by multiplied by 5.


The quality of competition numbers used in this calculation can be found here.



Only those players with at least 150 ES minutes played were included here

DEFENSEMEN

Quality of competition had a substantial impact on 5 Montreal Canadiens defensemen. Four of those players saw their even-strength risk/reward rating improve, while the other saw it drop.

Among defensemen, Josh Gorges faced the highest quality of competition. As such,Gorges saw his risk/reward rating go from 1.86 to 2.00. Raphael Diaz also saw his risk/reward rating jump; going from 1.86 to 1.99. Andrei Markov and Alexei Emelin were the two other d-men to see their ES ratings improve.

PK Subban produced the highest risk/reward rating among d-men before the adjustment of quality of competition as well as after, as he saw his rating drop just slightly from 2.30 to 2.29. Francis Bouillon was the only defenseman hurt by the inclusion of quality of competition into his risk/reward rating, as his dropped from 1.79 to 1.74; giving Bouillon the lowest risk/reward rating.

FORWARDS

Quality of competition had a substantial impact on 8 Montreal Canadiens forwards. Three of those players saw their even-strength risk/reward ratings improve, while 5 others saw them drop.

Among forwards, Tomas Plekanec's rating went from an already impressive 1.50 to 1.70. Rene Bourque's rating went from 0.83 to 1.01, while Brian Gionta's rating went from 1.27 to 1.44.  Eller (1.83 to 1.96) had the top rating among forwards prior to the inclusion of quality of competition numbers, as well as after their inclusion.

Ryan White and Jeff Halpern both saw their risk/reward ratings drop the most among forwards. The inclusion of Quality of competition saw White's rating go from 1.45 to 1.28, while Halpern's rating dropped from 1.86 to 1.71.

Max Pacioretty, Colby Armstrong, and David Desharnais also produced lower risk/reward ratings following the inclusion of quality of competition. Pacioretty's went from 1.74 to 1.68, Armstrong's dropped from 1.72 to 1.69, while Desharnais' went from 1.45 to 1.40.

Context is an important apsect of player evaluation. The inclusion of quality of competition numbers into the calculation of a player's risk/reward rating helps produce just that context.

No comments:

Post a Comment