We propose a contagion model to describe the evolution of the political voting trends in Brazil after the dictatorship from 1985 to nowadays. We consider a fully-connected population divided in two voting groups, left and right. Each group includes three kinds of agents, sensitives, inflexibles and radicals. While sensitives may shift their left or right voting, inflexibles and radicals do not. Excluding political interactions with radicals, we found sets of pair interaction parameters which reproduce all the voting outcomes of past presidential elections from 1989 till 2014 excluding 2018. At this stage, we found that adding an empty third voting group overlapping left and right, the populist voting group, can yield Bolsonaro 2018 victory. The initial filling of the populist group is triggered by the breaking of the interaction barrier between sensitives and radicals, which had prevailed for decades. Within each voting group, some of the interacting pairs sensitive/radical recast the sensitive into a populist. This status change from sensitive to populist allows the populist to interact with sensitives of both left and right, thus creating an additional source of filling of the populist group. The process is shown to yield the 2018 55% for Bolsonaro. From the 2018 distribution of votes, we evaluate the parameter changes which recover the recent 2021 poll yielding a victory of Lula against Bolsonaro at a score of 64/36. The magnitude of the parameter changes are very small highlighting the high volatility of what would be a vote under the current conditions. We also present an alternative explanation for the Bolsonaro victory in 2018 election, based on a single varying parameter without populists. The aim was to reproduce all elections outcomes (1989 -2018) and the 2021 poll. For 2018 election and 2021 poll, we compare the results of the two scenarios, with and without populists. Regarding the two possible scenarios for the 2018 Bolsonaro victory, we discuss about a hypothetic Lula/Bolsonaro 2022 second round voting and the requirements to yield either Lula or Bolsonaro victory in 2022. Both scenarios are feasible.