Significant reduction in file size from Camera Raw to DNG

Question

Hi,
I am currently testing the conversion of Leaf camera raw files into DNGs for a photographer's archive. I am hoping to convert all of the mos files to DNGs because Leaf Capture and the Leaf Raw Converter are not being updated and because the photographer wants to have an Adobe centered workflow. In my testing I discovered that converting mos files to DNGs through ACR 8.4 and LightRoom 5.4 resulted in a reduction of file size by nearly 50%. A 44.5MB mos file became a 23.6MB DNG. From what I've read only about 15-20% of the camera raw file should be lost and all of the data lost should be propietary.
Here-in lies my quesiton, is there any way that I can track or determine exactly what sort of compression is being done to the mos file and what information is or is not travelling in the conversion to DNG?
These are the settings I have used for converting raw files to DNGs:
ACR:
JPEG Preview: Medium Size
Embed fast load data
Don't use lossy compression
Preserve pixel counts
Don't embed original
LIGHTROOM 5.4:
Only Convert Raw files
Delete originals after successful conversion
File Extension DNG
Compatibility Camera Raw 7.1 and later
Jpeg Preview Medium Size
Embed Fast Load Data
Thanks!

Answer 1

50%? - I thought we were talking about 15-20%?
In my first post I questioned why I was seeing a reduction in file size of 50% when according to forums and articles I've read I should only be seing a 15-20% reduction in file size. I then wondered what data I might be losing, which you addressed.
Same as what? - what were the results.
I was referring to testing I preformed on camera raw files produced during different years (all mos). I converted all files with the same ACR and LR settings and found that the DNGs always reflected a 50% reduction in file size. This test suggests that any conversion issues is not necessarily related to how the camera raw files might have been differently built across years.
Adobe's raw data compression is touted by DNG zealots, but I haven't scrutinized it enough to corroborate or refute.., but my experience is that reduction is relatively marginal. All of this assumes original is also compressed - if uncompressed in original source, savings would be large.
The files I am dealing with are definitely uncompressed which could account for the large reduction in file size. I didn't realize until I posted to this thread that converting to a DNG results in a compression of the original image data. I understand that this compression is supposed to be lossless like a lossless compression to a tiff and thus result in no decrease in image quality or harm to the original image. I am baffled by how it is possible that any compression of a file (especially  by 50%) could not result in a loss of important data but I will accept that it is possible to have a truly lossless compression and that the size reduction I am seeing could be a result of all of the different processes a file undergoes that you have outlined.
I looked into the effects that backwards compatibility has on the conversion process which might interest you http://dpbestflow.org/DNG#backwards-compatibility
I also posted to luminous landscape's forums http://www.luminous-landscape.com/forum/index.php?topic=89101.new;topicseen#new
Although it wouldn't surprise me if the DNG conversion process tossed the xmp-like metadata, and kept the original stuff, but it would surprise me if it tossed the original stuff - but as I said before, I haven't scrutinized for completeness so I don't know.
I've done testing in which I converted .mos camera raw files with their sidecar xmps and without their sidecar xmps. My tests revealed that the DNG definitely carries over xmp metadata although it is not clear to me exactly how it is carried and if anything is lost.