A novel autonomous atmospheric entry guidance algorithm for a high elevation Mars landing is proposed. The bank profile is parameterized in terms of reversal times, modeled as Differential Algebra (DA) variables. The system states are propagated and exploited to obtain the DA representation of the objective and the constraints functions. The objective function is designed to maximize the final altitude still preserving control authority and horizontal precision. The DA representation of functions gives additional information about their sensitivity to variations of optimization variables. This additional information is exploited to find the desired optimal solution (if it exists) with a light computational effort, avoiding less robust processes.