Strange correlation between knock input and ignition advance (wheel horsepower)

A while back i did a little experiment. I studied data i collected with my stock Toyota 1.3 4EFE NA engine. I wanted to see the correlation between the knock sensor input (Phormula Knock KS-4) and the ignition advance. I sorted data by rpm and found out, that the knock input would rise as the ignition timing was advanced (as i somewhat expected) but then fall to a relatively low level as the engine hit its "sweet spot" where it would make max. torque/horsepower. With further advanced timing the knock signal would rise a little and with another step would spike to an - through headphones - audible knock. (see attachments for data below)

I thought the falling knock input level near MBT would be just a strange thing the little 4EFE was doing and this would just be an indication that the engine is not knock limited.

But then yesterday we had a lot of fun with a turbocharged 1.6 Miata and again, i gathered as much data i could. I am now very surprised to see that the knock signal behaves very similar to the experiment i did on the little NA engine. We advanced the timing in 2° steps up until 20° btdc. The power hardly increased at this point so we stopped. Also at this point, the max. knock signal sits at its lowest level. (see attachments for data below).

Anyone else ever seen the knock level to fall like that when the engine is kinda at its "happy spot"?

Is this just coincidence or could this be something to go by in connection with observing torque/hp gains?

For completeness some engine data:

Test engine 1: Toyota 4EFE completely stock with MS2 ECU

Test engine 2: '90 Mazda Miata 1.6 with Rods, Pistons, 8.8:1 CR, unknown Turbocharger at 14.5 PSI boost

It is common to see the mechanical noise level being reduced when timing advances, if it was formerly over-retarded. This said, no real knock present