Internal battery function (from another thread)
The meter is reset when the battery is at full charge. At this point the amount of electric charge accumulated by battery is roughly equal to its 100-hour rate usually measured in amper-hours. Physically this is a number of electrons displaced between plates. One amper-hour is equal to 2.225 x 1022 electrons. The meter then measures current in and out of the battery and accumulates these numbers over time. This way it is know exactly (with measurement errors of course) how many electrons moved between plates. The ratio of the number electrons left to the number of electrons at full charge gives you SoC.
Determining battery SOC in non-linear profiles is the subject of one of my pet projects. https://github.com/nsaspook/mbmc/tree/master/swm8722
The Victron is a very good meter when calibrated but as others have said take it's SOC reading with a large dose of salt until it's tracking is matched with good data. As someone once said: 'Trust but verify'.
That tells you the electrical charge moved by the electric field generated by the chemical reaction inside the battery but it only provides some of the information about the intricate non-linear electrochemical redox reactions inside the battery that really provides the information about true SOC. Reaction temperature, charge rates, relaxation recovery effect, level of charge, charge efficiency, Peukert effect and a host of other factors can completely invalidate charge in/out counting when you have non-steady charge/discharge rates (PV or wind profile and/or pulsed loads like motors).
Charge acceptace (end amps) should be the primary factor on when to float but I like to extend the cycle a bit when it's met before going into float.