Educational systems are complex adaptive systems (CAS). The macroeffects of an educational policy emerge from and depend on individual students' reactions to the policy. However, educational policymakers traditionally rely on equation-based models, which are deficient in reflecting the work of microbehaviors. Using inappropriate tools to make policies may be a reason why there were many unintended educational consequences in history. A proper methodology to design and analyze policies for complex educational systems is agent-based modeling (ABM). Grounded in the theories of CAS and computational irreducibility, ABM is capable of connecting microbehaviors with macropatterns. The purpose of this study was to contribute to the application of ABM in educational policy analysis by constructing an agent-based overlapping generations model with hypothesized inputs to qualitatively represent the environment of the Taipei School District. Four research questions explored the effects of Taipei's 2016 student-assignment mechanism and its free tuition policy on educational opportunity and school quality under different assumptions of students' school-choice strategies. The simulated outputs were analyzed using descriptive statistics and paired samples t tests. The findings, which could hardly be revealed by traditional models, showed that the effects were complex and depended on students' strategies along with the number of choices students were allowed to make; the assignment outcomes for elite students were robust to the mechanism, and the free tuition policy worsened school quality. Although exploratory, these findings can serve as hypotheses and a guide for Taipei's policymakers to collect empirical data in evaluating their 2016 mechanism and tuition policy.