You say the speakers where custom, as in DIY? Did you measure the impedance? No offense, but either you aren't considering a factor concerning the speaker's drivability or your bias is the cause for the audible difference (assuming none of the components where faulty).
Don't make me use my Goldilocks argument, you won't like it.
The speakers are really '77 vintage JBL L212. What I did was built new mirror image cabinets and replace the original XOs with a newer version of the same.
A technology JBL calls "Charged-Coupled". The only changes to the xo are the caps; replace all original cap sizes with two caps that are twice the size, used in series pairs, with the common point of each pair connected to a 9 volt battery, through a 6 megohm resistor.
The result is that the audio signal never crosses the dielectric zero point of the caps, thus eliminating the phase shift that would occur if the signal were to cross the zero point.
As for any bias, I had none at the time I bought the Denon, I had bought it for a pair of JBL Studio L890s, it seemed to do the job.
Then I fell into a steal of a deal for 5 PT800s, put them in place of 5/7 L212s. Then split up the 5 L212s to two different systems and gave the L890s to my daughter for a wedding present.
That's when I connected the mirrored pair to the Denon, and a TT. The resulting sound was less than I thought it should be.