Modeling the longitudinal dynamics of a fixedwing unmanned aerial vehicle (UAV) at high angles of attack is not an easy task. Indeed, when the airplane approaches stall, non-linear effects appear, including transient behaviors and an aerodynamic hysteresis. Although some models are present in the literature to address these aspects, they are usually aerodynamics-based and often too complex for analysis and control applications. Therefore, this paper presents a new hybrid polynomial formulation for the modeling of the aerodynamic coefficients. In addition, a Linearly Constrained Least Squares (LCLS) process guaranteeing continuity at the mode transitions is proposed for the identification of the model. The Hybrid Polynomial Stall Model (HPSM) is finally identified on experimental wind tunnel data, showcasing its ability to accurately predict a UAV's dynamics.