Context. In globular clusters (GCs), blue straggler stars (BSS) are heavier than the average star, so dynamical friction strongly affects them. The radial distribution of BSS, normalized to a reference population, appears bimodal in a fraction of Galactic GCs, with a density peak in the core, a prominent zone of avoidance at intermediate radii, and again higher density in the outskirts. The zone of avoidance appears to be located at larger radii the more relaxed the host cluster, acting as a sort of dynamical clock. Aims. We use a new method to compute the evolution of the BSS radial distribution under dynamical friction and diffusion. Methods. We evolve our BSS in the mean cluster potential under dynamical friction plus a random fluctuating force, solving the Langevin equation with the Mannella quasi symplectic scheme. This is a new simulation method that is much faster and simpler than direct N-body codes, but retains their main feature: diffusion powered by strong, if infrequent, kicks. Results. We compute the radial distribution of initially unsegregated BSS normalized to a reference population as a function of time. We trace the evolution of its minimum, corresponding to the zone of avoidance. We compare the evolution under kicks extracted from a Gaussian distribution to that obtained using a Holtsmark distribution. The latter is a fat-tailed distribution which correctly models the effects of close gravitational encounters. We find that the zone of avoidance moves outwards over time, as expected based on observations, only when using the Holtsmark distribution. Thus, the correct representation of near encounters is crucial to reproduce the dynamics of the system. Conclusions. We confirm and extend earlier results that showed how the dynamical clock indicator depends on dynamical friction and on effective diffusion powered by dynamical encounters. We demonstrated the high sensitivity of the clock to the details of the mechanism underlying diffusion, which may explain the difficulties in reproducing the motion of the zone of avoidance across different simulation methods.