I will present recent developments in analytical methods to predict abundance, clustering, velocities and bias of Dark Matter halos. In the standard analytical approach, halos are identified either with sufficiently high peaks of the initial matter density field, or with the largest spheres enclosing a sufficiently high density. I will revise the physical assumptions leading to this standard picture, and show how a careful statistical implementation of the model of collapse (even in the simple spherically symmetric case) leads to a surprisingly rich structure. This allows to make simple – yet remarkably accurate – analytical predictions for halo statistics, a necessary ingredient on the road to precision cosmology.