Based on the complex dynamical system perspective, approach and avoidance motivations in achievement context have been conceptual- ized by Gernigon et al. (2015) as two competing attractors. The strength of approach and avoidance attractors is assumed to evolve over time depending on the complex interactions among three key social cognitive variables (competence expectancies, expected benefit for the self, and threat for the self) that take place within and across personal, contextual, and situational levels. The complexity of these interactions is not accessible via conventional statistical tools, but can be modeled by computer programs (Gernigon et al., 2023) such as Agent-Based Models (ABMs). The first aim of this work was to develop a first version of an ABM capable of simulating the dynamics of approach and avoidance motivational patterns consistently with the literature on achievement motivation and with Gernigon et al.’s (2015) dynamical model of approach and avoidance motivation. The second aim was to compare the data resulting from the simulations with longitudinal data relating to the motivational states reported weekly by 10 athletes who were pursuing an important mid-term (from 1 to 2 years) goal. De- trended Fluctuation Analysis (DFA; Peng et al., 1993) was used to detect typical signatures of complex dynamical phenomena in the form of 1/f power-law distributions (i.e., pink noise) in the time series of both virtual and ecological data sets. The time series of the ecological data showed 1/f distributions, whereas those of data resulting from the ABM’s simulations did not (brown noise). In other words, the ecological data did reflect a non-linear dynamics typical of complex systems, whereas virtual data mainly evolved under the influence of random information. Therefore, future improvements to the ABM of approach and avoidance motivation are needed to make the model more able to resist to external perturbations, consistent with the nonlinear dynamics observed in real life.