Not to give you a cold shower
but you have to remember these are just numbers. They don't do you much good unless you compare identical pages claimed to be optimized the same way.
If your page has 2 images on it, your site doesn't call any external services, fonts, etc and the competitor site age has 20 images, uses FB, Twitter, etc - the comparison is like apple and oranges.
GTmetrix will not tell you all this and will not take it into account. Also you have to take into account the hosting and its optimization.
XC4 will always be faster than XC5 on the same server under same environment. It is the platform itself.
And lastly - none of these page speed services uses real user connection / device to test pages. They try to set it close to this but still not real people.
However - keep using them and try to follow as much as possible whatever makes sense for your website/platform to adjust and improve. Comparing to others is sort of useless but using it to improve your site virtual load times - it is good.
Sometimes they point out issues you won't see or encounter yourself or your dev.
The improvement you see with Phill's theme is because the XC stock one sucks. Qt after so many years and posts still does CSS extremely wrong and this has an affect on sites. XC CSS is so bloated it is a nightmare to use. Applies to both XC4 and XC5.
Anyway. Good job.