I hear what Burakkucat is saying. He is applying logic, a not unreasonable thing to do. My experience though leads me to question this, while having nothing to offer in its place. My ADSL2 modems always consistently run at a downstream SNRM (but not upstream!) that is far lower than the supposed target. It's as if the actual target that they are really going for is not that which we would expect, as if they were tweaking the given value just to be super aggressive and get faster sync rates by design. I don't know that this is the case. It would be very interesting to try and take an SNRM snapshot really soon after a resynch, but I've only just thought of this. I should do this to check the Burakkucat picture. Burakkucat has far more knowledge and experience than me, so be warned.
Some or all (I'd have to recheck) of my three DSL modems are set to a downstream target SNRM of 3dB, yet they run at 0.6 - 1.8 dB typically. If set to a 6 dB target they run at less than half of that. I don't know whether there is approx a fixed difference here ( target - d ) or a fixed ratio ( target * f ). They run really well even at these low d/s SNRM values. Upstream for some reason always sticks to the expected value, around the 6dB target. The latter fact doesn't exactly lend support for my tweak-by-design candidate theory: if you're going to cheat, why not do so upstream as well? One vague idea: that there was some loose thinking concerned with the popular practice by network operators of capping upstream sync rates, but that doesn't make sense to me.
And welcome to the forum too btw. I need to point out that I am using ADSL2 on extremely long lines, ~66dB downstream attenuation 7.3 km, 2600 - 2800kbps downstream sync and I have three supposedly identical lines that are bonded together to give a triple-speed pipe in both directions. This is handled by my ISP and my router.