Now that CMB observational data has increased in precision, the question of which (class of) models are the best candidates to realize inflation is particularly urging. Hybrid inflation is one of the three main classes of inflationary models and is considered very motivated among high energy physicists. I will discuss two of its main problems: the blue spectral index in disagreement with WMAP5 and the space of initial conditions for inflation to be successful, considered as fine-tuned. I will show that the slow roll approximation should not always be trusted to study the model, and discuss what we learn from the exact dynamics in terms of spectral index and ways to realize hybrid inflation. I will show that both problems mentioned above can be avoided by considering superplanckian initial values for the fields and discuss the consequences for the models.