All that counts in the end needs a contextual notation: in determining which team won the game. The very same statement is not true in terms of analysis.
Example:
1) You rank each of the 30 teams from 2007-2008 season to last in terms of goal differential, 5v5 goal differential, standings points, win%, or Corsi (all 5v5 shots: goals, saves, misses, blocks) for the first half of the season.
2) You take the same teams and rank them by goal differential (or win%) for the second half of the season.
Corsi from list number one will align closer to the order of list number two.
Why is that?
A smart person once wrote:
Source:
https://hockey-graphs.com/2016/11/0...-statistic-in-hockey-and-should-be-abolished/
We didn't start using these stats out of randomness or personal preference. It's because they were useful and tested better. When GF% and CF% tell you different things about a player or team, CF% was right more often than GF% (or you could think of it as wrong less often).
And this is important because that's how we continue to progress.
1) Corsi replaced goals as the go to metric, because it told you more about which player was better and was right more often.
2) Adjusted Corsi replaced Corsi as the go to metric, because you were able to reduce score effects, and it then told you more about which player was better and was right more often.
3) Expected goals replaced Adjusted Corsi as the go to metric, because you were able to introduce shot quality factors, and it then told you more about which player was better and was right more often.
4) XPM replaced Expected goals as the go to metric, because it adjusted for usage factors (like opponents, linemates, zone starts, coaches, and schedule), and it then told you more about which player was better and was right more often.
5) WAR "replaced" XPM as the go to metric, because it added other factors to XPM (scoring, shots penalties, face offs, etc.), and it then told you more about which player was better and was right more often.
The next thing will be right more often than the next. That's the scientific process.
I should also add, that it's good to go back along the different previous statistics when analyzing players that are similarly ranked in WAR and look at the individual inputs for more refined analysis.
Stats only tell you part of the picture. Not the whole thing. But, part of the picture is important and should not then be ignored.