Over the next little while we'll be bringing you an avalanche of analysis on the first half of the Canadiens season. Here is an introduction to the methodology we'll be using to evaluate the players.
Last season I began a project to review each Montreal Canadiens player every 20 or so games, or every quarter of an NHL season. This year because of the lockout we split the season in half at 24 games. Only players with 10 or more games will be examined, so while Tomas Kaberle just makes the cut, Yannick Weber, Gabriel Dumont, and Greg Pateryn don't.
There is one minor chance to the methodology from last year to this one, and that is along with the raw numbers associated with each statistic, we're going to list where each player ranks in each stat in comparison to his team. Defensemen will be ranked against each other, forwards against each other.
For this post what I want to do is explain each stat I'll be using, and what they bring to the table for analysis. As before, I'll be splitting stats up into even strength, powerplay and shorthanded situations. You can also look at our glossary of terms for advanced stats.
Thank you to Gabriel Desjardins of Behind the Net, Olivier Bouchard of En Attendant Les Nordiques (and EOTP), David Johnson of Hockey Analysis, and Christopher Boucher of Boucher Scouting for creating the websites and databases that allow us to provide these statistics.
Time on Ice [TOI/60] - will show how much a player is relied upon in any given situation. This will be displayed in minutes played per 60 minutes of game time. The more a player plays in any given situation, the more likely they're playing well, and relied on by the coaching staff.
Score Close Shot, Fenwick, Corsi Percentage - are all proxies for possession. While some would argue that possession during tied situations has less trouble with score effects, the increased sample size has its advantages as well. Last season we expressed these numbers as out of a factor of 1000, but this year we're just going to use a decimal and express the percentage to make it easier to understand at a glance.
Corsi Relative [Corsi Rel] - is a possession statistic from all of even strength play, relative to a player's teammates. Using this statistic allows us to eliminate the noise of team effects on possession. On a great team like the Los Angeles Kings, nearly everyone has positive possession, but not everyone is a great player. We can see who is strong and who is weak using this statistic. Expressed as a +/- number per 60 minutes of even strength play, for example if a player is -5, his team is outshot by 5 shots every 60 minutes he's on the ice.
Offensive Zone Start Percentage [Off Zone Start %] - allows us to see how a coach deploys his players. Who gets defensive responsibilities, who isn't trusted to play in their own zone, who is used in an exploitation role to generate offense? For zone starts, the toughest zone starts will rank highest, which will better show who is getting tough minutes.
Offensive Zone Finish Percentage [Off Zone Finish %] - shows which players push the play from their own zone into the opponent's zone. Finishing in the offensive zone more often than you start there is a positive, although ending your shifts in the offensive zone more than 50% of the time is always positive.
Corsi Relative Quality of Competition [Corsi Rel QoC] - using the weighted possession metrics of the opponents of each player, we can see the typical competition they face. The stat is by no means flawless, but tells us a lot about which players are used in a tough role, and is especially applicable to defensemen.
Corsi Relative Quality of Teammates [Corsi Rel QoT] - is a stat that we haven't used in the reviews before, but I've mentioned it a few times. It's the same idea as quality of competition, but measures the kind of support a player has from the players he's usually deployed with.
On Ice Shooting and Save Percentages [On Ice SH and SV%] - measure the average shooting% of the team while the player is on the ice, along with the average save% the player receives from the goaltenders he plays in front of. These numbers are very transient from season to season and large outliers from the norm are due to random variance that isn't likely to continue.
PDO - As discussed before, PDO is just on ice shooting and save percentages added together and multiplied by 10 to get an expression out of 1000. PDO trends close to 1000 for all players over time, although players can have slightly inflated or deflated PDO's as a team if they play in front of an excellent or terrible goaltender.
Scoring Chances For/Against - scoring chances are any shot or missed shot within the home plate area. They are recorded by our very own Olivier Bouchard on his site En Attendant Les Nordiques en Français. Scoring chances are a great way to measure a player's performance via generating offense as well as limiting offense from the opponent.
Scoring Chance Differential - is simply scoring chance +/- per 60 minutes of play.
Penalty Differential [Pen Diff] - Is a simple plus/minus score of penalties drawn vs penalties taken. A positive score is good, a negative one is bad. This is meant to show the impact a player can have in putting his team on the powerplay or playing disciplined.
Risk/Reward - is our own Christopher Boucher's stat from his website Boucher Scouting which analyses how many positive events a player is involved in per minute of ice time, minus how many negative events a player is involved in per minute. The higher the risk/reward number, the more effective the player is.
Goals, Assists, Points and Shots per 60 - is simply each stat per 60 minutes played at even strength.
True Plus/Minus - is a +/- statistic that eliminates the noise of inherently random events like empty net and shorthanded scoring. Only even strength play is counted.
Corsi, Corsi Quality of Competition, Corsi Quality of Teammates - is used for special teams because I don't think there are enough minutes played by each player to make relative statistics that significant. Also there are fewer lower end players who play on special teams, which makes relative statistics less significant.
Player grades are based on how a player is playing relative to the expectations of their position. For example, it would be a lot easier for a player like Travis Moen to get a high rating considering the limitations of a 4th liner than it would be for David Desharnais, who is expected to produce a lot of offense. In the end, all ratings are still subjective to the person giving them, so you're always free to give your own ratings in the comments and explain why.
Here is an explanation of what the ratings mean:
10/10 - This player is having a career year. It's incredibly unlikely that we could expect more from this player.
9/10 - This player is playing excellent, equal to the best they've ever played.
8/10 - This player is playing above what we would expect from a player in their role.
7/10 - This player is playing at league average for their role.
6/10 - This player is playing below league average for their role.
5/10 - This player is not fit to play this role.
4/10 - This player is an abject failure at the role he's expected to play.
3/10 - This player is actively causing the team to be significantly worse because of his play.
2/10 - This player is not good enough to be an NHL player for any team, in any role.
1/10 - This player is one of the worst players in the league, and likely would be one of the worst in the AHL.