This probably is impossible.
Which JK BMS is used? I have the JK Inverter BMS PB2A16S20P hardware version V15A.
Before updating to the last firmware the SOC alwaays was reported too high. After 2 weeks without reaching balancing voltage the BMS reported 50% SOC at 49 V. At 49 V the real SOC is about 10%.
With the new firmware this problem should be solved. And indeed, the reported SOC now seems quite accurate, perhaps even a little to low. But then a I noticed that JK achieved this by using a dirty trick: while - after calibration - the charging current is measured quite accurately, discharging current is reported (and probably used for SOC calculation) with a huge factor. I.e. the BMS reported 18 A discharge, but a reliable independent ampmeter says that there was only 12 A discharge.
Sorry to have to say that: you can’t rely on current and SOC data reported from the JK Inverter BMS. Calibration does not help.