In the Midwest we usually expect - 20F every winter, and chase the Iowa cold record (-47F) occasionally.
I have run the following tests, using the calibrated chamber of a major defense contractor!! (-40 degrees is a Military Spec, almost every electronic device in an airplane must be extensively tested at this temperature)
I took some clear plastic tubes and put a marble in each one, then filled them with oil. Yes, the marbles were precisely weighed - they are identical.
A piece of double sided tape and stuck them all on a board so they could all be turned over at once, and watch the marbles fall.
Wow, wish I had lot more samples but at - 40 degrees.....
Valvoline 30W .......Marble never moves, oil frozen solid.
valvoline 10W40 60 seconds to the bottom
Valvoline 10W30 45 seconds to the bottom
castrol 5W30 25 seconds to the bottom
Mobil 1 0w20 5 seconds to the bottom
Let me caution anyone doing a pour test - the samples heat rapidly, even a few seconds out of the cold chamber and everything changes. The last sample tested will perform better than the first sample. that is why I put them all on a board and turned them over together. If you turn the board over again, everything will be at least twice as fast! The last few degrees make a huge difference, and of course viscosity charts show that.
Only the Mobil 1 in the above sample was synthetic, a very incomplete test to say the least but... That is what was in my Escort last winter that successfully started at -40.7 last winter, none of the rest of the "outside" cars on my street could manage more than a low moan, let alone start!!
For a full report on that event:
https://bobistheoilguy.com/forums/posts/1343785/