Among the literature on self-regulated learning and social networking, the studies, which explore the impact of social networks on learning regarding connection sizes and relationship-establishing factors, are rarely seen in the context of social networking among strangers. This descriptive study addresses the gap by exploring data from 468 Chinese junior high school graduates in an online learning resource platform with an integrated social network. The data is digitally generated when the graduates engaged in online self-regulated learning activities for an average of 36 days without any facilitations. The data analysis explores the connection sizes and types of follow links, types of self-regulated learners, and their relationship with lesson completion. The study reveals that social networks trigger different levels of learning engagement. Specifically, the graduates with bidirectional follow links and the optimal connection size of five complete more lessons than other graduates. The study also finds that academic factors (similar learning goals and achievement gaps) are more important than social factors (common identity) in establishing social connections to support self-regulated learning activities. These findings have direct implications for the design of social networking that facilitates self-regulated learning, and enhances students' self-regulated learning efficacy in online learning environments.