The problem of choosing the optimal behavior of agents in the classic one-product model of the competitive market under linear functions of demand and costs of agents is considered. The dynamic decision-making processes under conditions of uncertainty of decision making by competitors, performed as repeatable static games within a range of admissible answers, are investigated. The analysis targets processes with rational agents using approaches of multi-step reflexive games and models of collective behavior to refine the solutions while observing the current market prices of goods. The processes are distinguished by choosing the current targets: in one case, the agents choose their current output as the current targets when refining the solutions; in another case, the current targets are perceptions of agents about the current equilibrium marginal cost of other agents. It is shown that for oligopolies with Cournot and/ or Stackelberg competition, processes with agents focused on the expected optimum output are more preferable than processes with agents focused on their perceptions of the equilibrium marginal costs of competitors because there are greater opportunities for finding equilibrium states.