Let me rephrase:
How do you calculate PDSCH SINR with that formula when there are 4 cells overlaping, each one with a different PCI, and their RSRP window of those 4 is between 4-5 dB?
RSRQ impacted by loading in cell.
Good threshold to shift traffic.
All.
Including overlapping, etc is taken care by RSRQ.
How come, if you don’t even know their PCIs and their mod3 value?
Again, I don’t know whether the formula is correct or not.
But is till now there is no solid reason why it can be wrong.
Why do we need to know that?
Interference already included in RSRQ (RSRP/RSSI).
RSSI included everything.
This looks logical.
Its not wrong.
But what I studied for RSRQ and SINR the relation is difficult to interpret for the reason I mentioned.
If RSRQ changes from -11 to -11.5 SINR might change from poor to good value and there will be sudden switch from QPSK to 64-QAM. 
I totally agree.
It’s having logarithmic relationship.
Yes.
2ndly UE measure RS power.
So it’s easy to find RS SINR and estimate PDSCH SINR.
Corrections can be made easily by incorporating other things like HARQ [Ack/Nack].
So ultimately we will have correct value of SINR.
Logically if PDSCH is 3 dB weaker than RS then there is no way it’s SINR would be better than RS SINR keeping in view constant neighbor impact.
Rest of the corrections are made by algorithm.
The target is to get accurate PDSCH SINR for DL AMC scheduling. The correct way is to using CRS in LTE or CSI RS in 5G to get the accurate signal power and noise and interference estimates to derive SINR then convert it to CQI to report.
Why using RS instead of PDSCH, this is because RS sequence has good properties for accurate average power and noise estimation, including low PAPR, good cross and auto correlation properties, stable and predictable patterns. But PDSCH includes random properties with random payload, whose accuracy is not reliable. If you tracking the LTE and NR standard, it is easily to see why to use RS to report CQI.