Glad to know that you have found my posts interesting and that I’m not the only slightly obsessive stats-head out there
I quickly discovered that trying to track the SOC of the PW2 by cumulatively tracking power in and out of it was, at best, approximate and you will quickly build up a cumulative error, even when factoring in round-trip efficiency (which for me has calculated to be about 93.5%). The efficiency does appear to vary over time, probably temperature related. Also my PW2 has lost a fair bit of capacity (more than 20% over 3 years) and is probably on track for warranty replacement, and that also confuses the picture.
I used spreadsheet formulas to cap the SOC at 10% (my minimum) and 100% and then start tracking anew from that point, i.e. when it was clear no more power was coming out of the battery then its SOC must be 10%. Equally when there is solar export the battery must be full. This only works if you are in self-consumption mode. If you are in TBC then forget it - the PW2 can do all sorts of weird stuff and you have no idea.
People have worked out how to extract all the historical PW2 data including SOC (which is only recorded every 15 minutes though, and only to the nearest percent) by writing code to access the Tesla API. It’s not straightforward and code regularly breaks when Tesla changes the API.
What my analysis has shown is that the payback time of the PW2 in my case is not great - more than 10 years - and additional PW2s would have exponentially longer payback times (> 20 years). But I did not buy it for economic reasons - I had a personal desire to maximise self-consumption and minimise grid usage, just because.
But as my post above shows, the more grid prices go up, the payback time of the PW2 reduces