Buildings can be operated in an energy-flexible manner while respecting occupant thermal comfort. This energy flexibility of building operations, both in time and quantity, can be harnessed by the electrical grid for load balancing. In the context of smart grid and intelligent buildings, the concept of energy flexibility in buildings broadens the existing demand management thinking from the top-down one-way control to two-way communications. This paper, extending studies on thermostat controls of heating and air conditioning systems for demand response, evaluates the impact of different control schemes on the energy flexibility of residential buildings. Two control strategies, Model Predictive Control (MPC) and Rule-Based Control (RBC), are investigated for a space heating system using co-simulation studies. Four indicators are introduced and adapted from the literature to assess the control performances of the strategies. Simulation results show that different flexibility indicators favour different control strategies in this case study. For demand response events of four hours, the MPC strategy presents two to three times of flexible energy than that of RBC. MPC also delivers 20％ more of maximum power reduction during the events against RBC. The RBC strategy, on the other hand, is twice of MPC for flexible energy efficiency. This evaluation work can be beneficial to guide the control system design of new buildings or control retrofits of existing buildings that consider better grid-building interactions for the future.