Modern developments in Shape Memory Alloys (SMA) has positioned the material as an attractive alternative actuation for high yield, low cost industries which stand to benefit from the materials simple form, light weight, and high energy densities. However, the speed and predictability still remain as a barrier to its acceptance and usage. The robotics community has shown promising results with antagonistic actuation architectures to increase the cyclic speed and produce controlled motions; however, such control-based approaches generally require sensing and feedback implementations and tuning that are undesirable for high production products. This paper presents a simple but effective physically-based thermodynamic model for generic antagonistic actuation architecture. The model is derived from three sets of equations: differential equations describing the thermomechanical phase transformation behavior of the material, compatibility equations specific to the antagonistic configuration relating stresses and strains in the two wires to each other, and heat transfer equations involving the thermal properties of both the environment and the wire material. This model takes into consideration several key-aspects of real devices such as the wires becoming slack or localalized boiling conditions. This model was experimentally validated and studied under a range of conditions including variations in driving frequency (0.3-10 Hz), duty cycle (10%-45%), amplitude (50%-100% transformation), and wire diameter (8-20 mil). The correlation over these widely varying conditions indicates the model's accuracy and potential for use in the design process of future antagonistic actuators and their controllers for industrial applications.