Suffice it to say, when someone asks how a team got where it got when they have poor public-data model analytics, the microstats show you how it happened. CF% and xGF% show you what happened, which is a totally different question.
I will tell you this though. Because Valiquette is also an intermission analyst on MSG during Rangers games, he also often shares game reports produced by CSA. Before anyone says his connection makes him biased, CSA itself isn't and they have NHL teams as clients (not the Rangers). Those game reports are very often better for the Rangers than, for example, NST's reports for the same game. There's a common theme among the CSA reports. High % chances the Rangers are often equal or better than opponents. Medium % chances, the Rangers usually have the edge, sometimes by a lot. Low % chances, the Rangers often get killed.
That explains a lot about why the public models, which basically only care about shot volume and location, are worse for the Rangers than the CSA model which accounts for pre-shot movement, screens, etc. Context is important.