In this work, we address the description of the dynamics of cyclodextrins in relation with nuclear magnetic resonance (NMR) relaxation data collected for hydroxymethyl groups. We define an integrated computational approach based on the definition and parametrization of a stochastic equation able to describe the relevant degrees of freedom affecting the NMR observables. The computational protocol merges molecular dynamics simulations and hydrodynamics approaches for the evaluation of most of the molecular parameters entering the stochastic description of the system. We apply the method to the interpretation of the C-13 NMR relaxation of the -CH2OH group of cyclodextrins. We use gamma-cyclodextrin as a case study. Results are in agreement with quantitative and qualitative analyses performed in the past with simpler models and molecular dynamics simulations. The element of novelty in our approach is in the treatment of the coupling of the relevant internal (glucopyranose ring twisting/tilting and hydroxymethyl group jumps) and global (molecular tumbling) degrees of freedom.