On the topic that usage does not matter at all: Let me illustrate one -- of MANY -- problems with the way the measuring of QoC is done.
Against Dallas Pionk played about 72% of his 5 on 5 ice time against Seguin, Rads and Benn. This is the shift chart from that game:
View attachment 157383
As you can see on 24 occasions there is an overlap
at either the very start or the very end of a shift (remember this because it will be relevant at the end). So of the ice time Seguin, Radulov and Benn has against other Rangers than Pionk, a significant part of those 2-3 minutes is the very first seconds of a shift or the very last seconds of a shift. Everyone can see that right?
If shots were evenly distributed during a shift this would not impact the numbers. I.e. if as many shots was taken by a line on average like the 1st second after a shift started, the 3rd second after the shift started, the 5th second after a shift started, the 7th second after a shift started and so forth. So is this the case?
To me its not a surprise that this is not the case. This is how the shots are distributed in relationship to the time passed since the shift started.
View attachment 157387
As you can see there is of course a dramatic increase in shots taken the first 8-10 seconds of a shift. The biggest difference for shots against is of course the first seconds after a FO, and that isn't relevant as far as I can judge since you get the match up you want from the time the puck is dropped after a FO. But there is a dramatic increase in shots against after an on the fly shift during the first 8-10 seconds. In addition we can see how S/60 levels off the longer into the shift you get.
So what we can see is that there is a Golden Age of a shift for an offensive player. It starts about 5 seconds in and peaks at about 40-50 seconds.
So when do you want to play against a top player to get top CF%?
1. Early on a shift.
2. Late on a shift.
3. Absolutely not the first 5 seconds of a shift after a FO in your own end.
So lets apply this to the actual shift chart above: It is obvious that Pionk faces a larger portion of Seguin, Benn and Radulov's Golden Age of a shift, ie there most productive time of a shift, and that other players on our team instead faces a much larger portion of the most ineffective time of Seguin, Benn and Radulov's shift.
In addition:
1. The above is -- of course -- true. I support it with data, but that is of course unnecessary because anyone with half a brain can figure out that this is the case. Its extremely logical. Is anyone really surprised? Didn't think so.
2. Why is so many "experts" on this topic completely unaware of the above? Well they can answer for themselves (OK we know that they are world champions of putting their heads in the sand and that we won get a reply, but still...), but one HUGE reason that is very obvious when you read articles, follow ppl on twitter and what not is that NOBODY IS LOOKING. Everyone active with metrics in public channels is looking to confirm how reliable the numbers are, nobody -- that at least haven't come up with a solution for a flaw -- is looking for them.
3. This is just one of many many many flaws in the numbers. Another very big factor that I am 110% sure of that I can prove is the existence of real
Poison Pill-shifts. You get 15-20 CAs against during a game. It is very common during a game that there are a number of shifts that are very predictable that they will mean a lot of CAs against during that single shift.
The most obvious example is of course when you are up by a goal with 50 seconds left and a FO in your own end. The stats those last 50 seconds in that situation is extreme. But there are other shifts too that most certainly breaks the pattern. You want to bet that the D the coach trusts the most get the by far biggest burden of those shifts too? We are probably talking 90% of that ice time. If you ever coached, how easy isn't it to recognize the feeling of "this is going to be a tough shift against?" And you basically never spend a ton of time in the attacking zone those shifts... And there are many factors like that.