With the odd exception, I generally write beat/bass heavy electronic music at a variety of tempo.
I don't use anything on my Master channel strip whatsoever, other than a single metering plugin; specifically Sonalksis Free G which I have found to be the most useful/accurate of those I've tried. The volume dial on my monitors provides boost if required to reference my work to other mastered tracks or sounds, but I don't use a plugin chain for this purpose.
Any problems that arise are addressed in the mix on a track by track basis, and never via the Master bus.
If I have rogue peaks or frequency issues, I isolate the sound(s) responsible and control it (them) in a manner suitable to the source: various forms of compression, clip envelopes to manually control parameters and EQ would commonly feature for this purpose, though the specifics would depend heavily on the source sound and the cause of the problem.
Although I don't separate my arrangement/mixing stage as distinctly as memes_33 has suggested, generally speaking I subscribe to much the same approach, particularly comments like this one:
----------memes_33 wrote:personally, i think compressors are something that should be applied as-needed, rather than something that you automatically put on tracks as a matter of course
So to take memes_33's comment a step further, if you're having problems with your mixes, rather than adding stuff to the Master bus to fix them, it might be worth applying the same concept to your programming, recording and/or arrangement.
In my humble opinion, this is an often unmentioned, but essential part of achieving sonic balance.
Thinking about not just how or what, but when sounds should play is as much about the balance of your mix as it is about realising a more creative aspect of your work (ie. groove/harmony etc.). In short, wise choices can help address all sorts of common issues and make the mixing process a far simpler task.
To elaborate with a somewhat overblown example:
It's all very well to stack fifty different audio samples to sound on the 'one' and mark the start of a breakdown with a monstrous hit for dramatic effect, but are all those sounds required to hit at identical times? More importantly, are all those sounds even required to get the same results as you might with far less? Of course it's possible that the answer to both questions might be 'yes', but it's easy to lose sight of the idea that production is very much about contrast, and sounds/silences are most relevant to the other sounds/silences that surround them.
In my own work at least, I find that problems such as rogue peaks or over-represented frequencies are often caused by lazy programming/recording/arranging; and as such, once located can easily be fixed without additional effects (especially on the Master bus).
Apologies if you find any of this post a little tangential, just thought I'd throw a different angle into the mix as this has been an interesting thread.
