For the record, YV* is not a colourspace, it's a shorthand way of writing the chroma subsampling ratio for planar colourspaced video. 420 will always be smaller than 422 which will be smaller than 444. Exact ratios are 1.5 bytes per pixel in YV12, 2 bpp in YUY2, and 3 bpp without subsampling. From that you can tell that your YV12 video should be half of the size of your fully sampled video, or 3/4 of the size of your YUY2.
Good way to know if your lossless codec configuration was done properly in comparisons too. Just sayin'.
I was experimenting on the 9th episode of Ore no Imouto (24mn 30s, 1280x720).
By rendering here I mean rendering a 30 sec 1-layer full-quality preview in AE.
Lagarith is slow but small. HuffYUV is a bit faster than Ut and even smaller if compared to speed-optimized Ut. And it is way faster in encoding.