For those that have been reading the blog for a while, you notice that I don’t shy away from using #fancystats at all. This post is aimed at explaining the metrics we use, what they measure, and why we use them. At first I focused mainly on GVT, and my search eventually led me to create PVT, both of which were discussed in detail in our Metrics We Use page.
While GVT and PVT are good metrics for determining overall value, defensive value is something that is a bit harder to quantify. With offense, you have the typical scoring stats that assist in determining how successful a player is. Defense doesn’t have that. The plus/minus stat is one of the most flawed stats in sports, as it relies heavily on ice time. Not all players are created equal when it comes to ice time. Some players get tougher minutes than others against tougher competition.
Quality of Competition (QoC) was the metric born from this conundrum, and is a fairly easy stat to understand. The more time seen against opponents top players, the higher the QoC. The value rangers from +1 to -1, which makes it a little difficult to note large differences between players. That said, the players with higher QoC are facing tougher competition than those playing with a negative QoC.
But defensive prowess isn’t just about the quality of competition faced, it’s about how successful a player is while facing that competition. Is this player facing tough competition and getting slaughtered? Or is this player taking this challenge and succeeded?
Corsi is the metric that we use to track this success. Corsi in itself is also relatively easy to understand: It’s just a count of the number of shot attempts (blocked, missed, saved, goals) taken at the opposition’s while a player is on the ice minus the number of shot attempts taken at the team’s net while the player is on the ice. The theory here is that a player is driving puck possession if his team is taking more shots than defending. If the number is positive, it means the team spends more time in the offensive zone while the player is on the ice.
In essence, Corsi is just the plus/minus of shot attempts. But the problem here is that time on ice isn’t a factor in counting stats, which is why I have an aversion to the raw plus/minus stat itself. So generally, you will see Corsi displayed as Corsi per 60 minutes of play, or Corsi/60 (also referred to as CorsiON, or Corsi per 60 minutes while on the ice). This eliminates the time on ice variable, and creates an equal playing field for all skaters. The 60 minutes comes from the full 60 minutes in your standard hockey game.
Relative Corsi (RCorsi) takes Corsi one step further, and compares it to the success of other teammates. It takes the Corsi of a player while on the ice, and subtracts the Corsi of the team while he is off the ice. The overall result here is how effective the player is at driving puck possession when compared to the rest of his teammates. The measurement of the stat is still the same, as a positive number represents a player who is driving puck possession, and a negative number is a player who is defending more often than not.
These two stats require more than blind stat reading. Taking one alone does not paint an accurate picture. For example, if you look at Dan Girardi and Ryan McDonagh, these two defensemen will have negative RCorsi because they play in their own end a lot. If you look at RCorsi alone, you would assume that Girardi and McDonagh just aren’t good at driving puck possession.
This is why you look at QoC in addition to RCorsi, and it gives you a completely different picture. Girardi and McDonagh face incredibly tough competition night in and night out, so it’s expected that they would have a negative CorsiON or RCorsi. Remember, RCorsi is in comparison to the rest of the team, which does not face the type of competition that McDonagh and Girardi face.
The last metric I like to look at is how the coach deploys his players after a whistle. I say after a whistle because that is when a coach really has the opportunity to match up based on face off location. In the offensive zone, he would want his offensive players out there (or some players he doesn’t trust as much). In the defensive zone, he would deploy those who he trusts in those defensive situations.
Offensive Zone Start is the stat that we use to measure this. This is also very simple, as it is simply the percentage of starts a player takes in the offensive zone after a whistle. If a player starts 7 out of 10 shifts in the offensive zone, his OZone Start would be 70%. I like OZone start because it shows how much faith a coach has in his player in his own end. While you generally expect your players to be at around 50% OZone start, there are a few exceptions to the rule.
Top line players –your go-to guys for offense– generally have 55% or higher OZone start. This makes perfect sense, as a coach wants to deploy his offensive players in a position where they are more easily able to capitalize and generate offensive chances.
Defensive shutdown players are the ones likely to have an OZone start under 50%, sometimes much lower. Brian Boyle is the perfect example here, as last season his OZone Start was a whopping 30%. That means he started 70% (!!!) of his shifts after a whistle in the defensive zone. That shows how much Torts relies on Boyle in those situations.
OZone starts aren’t necessarily about how good a player is in their own zone, but more about how much the coach relies on them in those tough spots.
I hope this sheds some light on some of the metrics I use here. As I begin to use others, I will be sure to explain them in detail. Also, I will create a permanent link to this in our Metrics We Use page for easy reference.
More About:Analysis