I think the best thing you can do apart from listening to that Mastering Show podcast about streaming service loudness normalization is to import audio from a CD that you want use as a reference to aim for and import a few songs in to WaveLab and analyze their loudness. You could also track down true WAV versions from HD Tracks website but just know that using an mp3 will give you inaccurate peak level readings because the peak levels increase when a WAV is converted to mp3 as well as other potentially bad artifacts. For comparison sake, compare WAV to WAV, especially the peak levels.
Also be aware that some even big name albums follow bad practices of too much loudness for the material and not leaving enough peak headroom for lossy encoding for streaming services and online retailers who sell data compressed formats. You have to think about if you primarily care about the sound for the CD release, online streaming release, or both. I assume everything I master will be on streaming though some clients still make CDs so I find a sweet spot that works for both in most cases.
I would forget about RMS and use LUFS Integrated and Short-Term loudness for measurement purposes. RMS is a dated and somewhat flawed measurement. LUFS is more in line with how our ears perceive loudness.
-5 RMS can maybe work on an already great mix for EDM and hip-hop but usually anything with real instruments will sound destroyed at that level. Especially if you have hidden low end issues due to monitoring, a mix can reach -5 RMS easily but still sound quiet and weak.
I’m not sure where you got that -5 RMS number but I would never advise it.
Just analyze loudness on the LUFS meters in WaveLab and use your ears to see what some of your reference songs are doing and aim for that. Depending if it’s the original CD master from early 1990’s or late 1980s, or a circa 2000s master, or a very recent streaming optimized remaster, you may find varying levels of the same album. We went from quieter CDs when CDs were new, to OK levels for a brief moment, then way too loud from 1996-ish to 2010-ish, and now the trend is slowly coming back to a sane level though some people still push it too far.
There is no way to suggest an ideal loudness level for your material without hearing it and even then it’s subjective but just know that at some point you will surpass the loudness potential of the mix itself and going any louder that that will just make it feel smaller, and going louder than that can just destroy it and make it unlistenable. Genre/taste pending.
I agree that compressing a mix just for the sake of making it louder before mastering is not a good idea. I would only use the compression on the mix before mastering if you feel it needs it for character and glue, or to control the dynamic range, but then I would also be careful about the make-up gain that it’s not driving it into the digital ceiling. It’s easy to push a compressor output loud into the digital ceiling and not realize you are essentially peak-limiting it already.
To preserve any potential peaks over 0dB in your mix before mastering you can save the mix from Cubase as 32-bit floating point and this way, turning down in WaveLab before Master Rig could be helpful but turning down something that has already hit a brick wall and then turning it up again often sounds over-processed very easily. That being said, some plugins have a hard ceiling at 0dB and some do not.