This article proposes a procedure for the control of electric vehicle (EV) batteries, aiming to have an optimal matching between local renewable production, domestic loads, and EV consumption. The procedure starts with the analysis of historical photovoltaic (PV), EV, and domestic load profiles. Load and PV profiles are forecasted using statistical-based algorithms, while the expected patterns of EV usages are forecasted using a combination of statistics and clustering techniques. Then, the forecasted profiles are used to estimate future energy balances trough an optimization process. Finally, the real-time management corrects the forecasting logic and checks the parameters of the EV storage to guarantee its correct and safe operation. Three different EV usage profiles (obtained by the clustering of 215 real users) are shown and their impact on the energy balance of EV–PV–home systems is quantified. The results are finally compared with those obtained with a traditional rule-based logic working without forecasts, by also reporting a detailed analysis of the main aspects having an impact on the results.