State of charge ≠ anode or cathode stoichiometry, but a reduced range. On the picture below, the bars represent theoretically available stoichiometric ranges of anode (negative electrode, NE) and cathode (positive electrode, PE) respectively:

Image from ‣

Image from ‣

On this chart, the bars are aligned so that any horizontal line which crosses both bars represents a possible stoichiometric state of the cell: the stoichiometry on anode increases from the bottom to the top, and the stoichiometry of cathode increases from the top to the bottom.

In some experiments, cycling on different reduced SoC ranges (e. g. 0–80% vs. 20–100%) makes cell to degrade with different speeds. This effect is contingent on the alignment of the anode and cathode stoichiometric ranges (similar to the picture above) in a particular cell type and to the cutoff voltages specified by the manufacturer, so the results can be very different for a different cell type. Another possible (albeit small) contributing factor for why cycling at the 20–100% SoC range might degrade cells faster is because the average SoC is higher, which contributes to faster growth of SEI layer on anode.

Coulomb counting shouldn't be used to estimate SoC because of unknown sensor bias and non-linear error, unknown (precisely) Coulombic efficiency (depending on temperature, SoH, current), unknown full capacity of the negative electrode (depending on SoH), self-discharge current, leakage (to voltage sensors) current.

Pack state-of-charge doesn't make sense. For "range" estimation, the remaining energy is important, not SoC. Average SoC across all cells (assuming they are close enough) could be used for setpoints to maintain margins in automotive applications.

Depend on the state of charge:

Ways to estimate state of charge: