Try this experiment as a warm-up:
Remember the components of the skies we made in class? There were two layers of grey (light and middle value), some blue and some white. We had clouds in mind when we painted yesterday, but imagine if you could put those brushstrokes in a big shaker, shake it all up and toss the marks arbitrarily onto the paper. I think it would still be recognizable as a sky once you put a strip of land below it.
To test this theory, wet a few places at the top of the paper, leaving the rest dry. Now make some strokes and washes of the colors like the greys and blue and white from class without regard for what goes where. You could throw a little very pale warm in there, too.
If we want to keep this experiment strictly scientific, it's important not to correct your sky. The idea is to find out how broad the guidelines are for this very common subject, and what the role of the context is in making the sky believable.
Now add the context. Even just one stripe of land will be enough to see if your random mess resolves into a sky, but feel free to put in some other familiar landscape elements; Hills, trees, bushes. It helps to make something tall enough to stick up in front of the sky, like a tall tree, a steeple or a fence post. Stringing some phone wires can be very convincing. This should all take just a few minutes (it's the correcting that takes the most time).
If you have time, make some more skies. Please keep the fiddling to a minimum. How else would you discover the real range of what works?
The purpose here is to stay mindful of the viewer's willingness to meet us halfway. Everyone likes to have a part to play in interpreting a subject.
Here are some paintings and photos of skies that may stretch your idea of what kind of shapes and colors will be acceptable: