The MARCS model simulates three statistical moments of annual runoff based on a mean annual precipitation for any projected period period of 20-30 years in the future. To run the MARCS model, non-central moments of annual runoff are calculated from historical time series observed in a gauging site for a period in the past considered as a reference period. For the reference period, a mean annual precipitation is calculated from observations, and it is expressed in mm year-1. For the projected period, a mean of annual precipitation is calculated from output of any global/regional climate model, and it is expressed in mm year-1. The MARCS model simulates non-central moments as well as mean value of annual runoff, coefficients of variation (CV) and coefficients of skewness (CS) for the projected period considered. The code of the model core (model_core.py) is supported by two files, which were used in an example embedded to the code. For the reference period, the non-central moments of annual runoff were calculated from observed time series extracted from the Global Runoff Data Center, GRDC, 56068 Koblenz, Germany (6233410.mon), and mean of precipitation was calculated from the dataset of NOAA/OAR/ESRL PSD, Boulder, Colorado, USA at a grid node nearest to the watershed centroid (6233410_pre_obs.txt).