I would do some surgical eqing. I pick a main supersaw that will dominate the pswctrum then layer very thin ones that would fill gaps. Then would make them really wide.

Well i make 1 chanel for the most "supersaw" i have, i make this to dont have much diference and all work togheter in harmony, i try to cut that area eq in 65Hz with a high gain, and the 150 Hz, or 200 Hz... and a short cut on low, this is work good to me because, i dont cut all low eq but i cut the line when i use in the bassline, and the supersaw need a bit of low frequencys to dont be so much high!

Take a look on this image, i cut in the fab filter that somethings area when i say for you, and in the channel i make more changes in the high, and med frequencys...

well i add some reverb in this channel but i make a individual reverb chanel to the other elements of the song!

Normally I grab about 5 saws (give or take, depends how fat I want it) and play around with them. Maybe I detune one that has 8 voices, then I leave 1 with only one voice. Maybe I take one multivoiced saw and get it one octave higher. Add a little bit of white noise, add reverb if necessary. It all depends on how do you want to use it and how it fits in the track.

After that I normally sculpt with EQ each layer to clean it up.

I normally make 2 or 3 layers, and I usually gave them their own space in an octave then I cut all the innecesary low end, and do some EQing depending on the characteristics of the layers and depending on how I want that particular supersaw to sound (very detuned, crisp and distorted, flat) and finally I do some compression to tight up the layers, little bit of saturation/distortion and some adjustments on the 200-500Hz area.

That's how I usually do it, but I would love to hear different aproaches on this, since most of the sounds are based on saws/supersaws.

Pretty much this. Have one sound carry the main body of the supersaw patch, then use layers to adjust it's timbre, tone, and brightness. So, having thinned out, mixed in layers that sit an octave higher, or are more detuned and stereo-spread, or have a different tonality (such as square waves, which are actually a really good supersaw accompaniment). Just add parts to improve what you've already got/fill in the spectrum according to your objective.

Here are some tips that I posted in a different thread but I thought might be useful here:

1. When building layered stacks, use different instruments to play upper and lower harmonies. Using the same MIDI chords for all layers can muddy up the sound.

2. The chord notes/velocities are just as important as the sounds/instruments.  Sometimes dissonant harmonies add a very nice character, but it may or may not work with a specific sound/instrument.  Try changing velocities of "problematic" notes or removing them completely if it does not work.  Conversely, if a chord sounds too thin, adding more notes may work better than trying to fatten up the sound.

3. The bass sounds/layers are very important.  It's usually a good idea to group/process the bass layers separately from the supersaw.  Play around with the bass melody (add extra note transitions, make some notes longer than the chords, play bass notes where the chords are not playing, etc). This will make it more interesting.

4. When building layered supersaw stacks, getting the top end right is probably the hardest part (for me at least). Supersaws tend to be harsh in the high frequency domain, so cutting them off and adding white noise may help. Also, try adding a square layer - they tend to have a pleasant glow.

5. Ambience is very important. Adding reverb and vocal fills in the background can make a huge difference. When using a background reverb, don't forget to process it (saturation and EQ helps it stand out and sit better in the mix). While the usual jumping reverb trick works very well, sometimes it's better to add a wide supersaw layer in the background with a long release. Reverbs tend to sound muddy.

6. Increasing the velocity of the top notes may emphasize the lead melody. Alternatively, add a layer that only plays the topline melody. Try changing up the notes of the topline melody to make it sound more interesting.

7. Don't overdo it. Check every layer to make sure it improves the overall sound.

8. Make sure to adjust the levels of individual layers. Sometimes a layer playing at a low volume can add texture and thickness, but can really mess up the mix if it's too loud.

Hope this helps!

I feel like people are overcomplicating making and layering saws a lot.
What I do is layer 3-4 sylenth patches with all variaties off undetune saws and detunede saws and perhaps, 1 oscillator of a retrig saw. I apply some saturation to the invidual layers, maybe Maserati. Then a lowcut between 200-350 hz depending on the rest of the track. Then route to a bus, add delay, reverb and maybe sausage fattener. Then I take out the highs above 10k with a 6-12 db slope and replace it with whitenoise.

I always add a channel with a a heavily detune saw with 100% wet reverb with a short tail. For me this add a huge special sound without a long messy tail (which losses the clarity of a song)

This is good idea.
Another way is to bounce your layers to audio and manually cut out the reverb or shorten it where you don't want it. I do this and feel it makes the track much cleaner in the end.