What is risk/reward?
Risk/reward rating is a data-generated measurement I invented in order to help determine how well a player is playing. It quantifies a player`s impact on the game by tracking successful and unsuccessful puck-possession events. Events used in the calculation include (among other things); puck-battles, loose-puck recoveries, passes, dekes, dump-ins, dump-outs, intercepted passes, pass-receptions, blocked shots, deflections, off-sides, and attempted shots-on-net. Each of these events is entered into a database as either successful or unsuccessful.
Risk/reward combines this data with ice-time to determine how many more successful plays than unsuccessful plays a player contributes per-minute of ice-time.
(Successful plays / minutes played) - (Unsuccessful plays / minutes played) = Risk/reward
Introducing ice-time into the calculation allows us to reward players who are more involved in the play. Simply tracking the ratio between the number of successful plays produced for every 1 unsuccessful play does not provide us with information about how involved a player is.
Calculating ratio only, Player A; who has made three successful plays for every one unsuccessful play would have the same rating as a Player B; who has made six successful plays for every two unsuccessful plays.
Breaking this data down by minutes played, and assuming each player has produced these results over the course of just 1 minute of ice-time, player A would have a risk/reward of 2.00, while player B would have a risk/reward of 4.00.
What does risk/reward tell us?
Risk/reward ratings can be broken down by zone (offensive, defensive, and neutral), and can be calculated for players at even-strength, short-handed, or while on the powerplay. For example, a player`s offensive-zone risk/reward rating would be calculated by using only those events that take place in the offensive-zone, while a player`s powerplay risk/reward would only use those plays that occurred while his team was on the powerplay.
Risk/reward rating tracks the play of each individual player. As such, it is a measurement of an individual player`s play, and does not need to take into account quality of teammates. That said, I have introduced Relative Corsi Quality of Competition numbers into the calculation with interesting results.
The sheer volume of events tracked (an average of 1,200 per-game for each team tracked) produces an incredible amount of puck-possession data. From this data we can establish (these are just a few examples of many); a player`s overall success rate when attempting to maintain puck-possession by way of a pass or deke, their success rate when attempting to remove puck-possession from the opposition by way of a stick or body check, or even their success rate when attempting to dump the puck deep into the offensive-zone.
The system I`ve developed, as well as the possible applications of the data generated, are constantly evolving. The experience I`ve gained through the countless hours of player-data tracking has allowed me to create detailed work instructions for the system in order to ensure objectivity through quality-control. One game can be tracked for an entire team in approximately three hours.