Current DESS algorithms sell and buy at absolute maximum/minimum prices, but that very often ends up not optimal.
When evening price is high and DESS estimates there is not enough solar to reach 100% battery, it takes missing energy from the grid by finding the least expensive time to do it. Often that ends up somewhere mid day and if weather changes and the sun comes out in the evening, DESS ends up with full battery and excess solar. This could be avoided if DESS scheduled charging later in the evening. Many times later price is just a fraction of a cent more expensive than mid day. This could be implemented by rounding prices a bit and considering insignificant price differences to be the same.
Here’s an example:
DESS sees that it will be missing a bit of energy and schedules charging at 13:45. But the price at 17:00 is just 0.001 EUR more, so it did not select that timeslot. Scheduling charging as late as possible would account for weather changes and maybe it will not be necessary to charge from the grid after all. I have observed situations when many kWh are taken from the grid mid day just to end up with excess solar in the evening when sell price is below zero.
Another benefit would be that charge and discharge could be distributed evenly across a longer time, not just full power when the price is highest (discharging) or lowest (charging). The charger could then work at peak efficiency rather than at full power. So that fraction of a cent less price is offset by lower charger losses.