This study investigates how team assembly mechanisms influence the structure of collaboration networks and team performance in creative fields. The research focuses on both artistic and scientific domains, analyzing data from the Broadway musical industry (BMI) and four scientific disciplines: social psychology, economics, ecology, and astronomy. The study proposes a model for team self-assembly based on three parameters: team size (m), the probability of selecting incumbents (p), and the tendency of incumbents to repeat previous collaborations (q). The model predicts two phases in team assembly: one where agents form a large connected cluster and another where agents form isolated clusters. The results show that team assembly mechanisms determine both the structure of collaboration networks and team performance. The study finds that team size and composition depend on the complexity of the creative task. In the BMI, team size increased over time as the complexity of the musical form evolved. Similarly, in scientific collaborations, team size increased over time, with faster growth in ecology and astronomy. The analysis of team size cannot capture the fact that teams are embedded in a larger network, which acts as a storehouse for knowledge. The structure of the largest cluster in the network is influenced by the distribution of different types of links, with a higher proportion of repeat incumbent-incumbent links leading to a more homogeneous knowledge pool. The study also explores how team assembly mechanisms affect the performance of teams. It finds that successful teams have a higher fraction of incumbents, who contribute expertise and know-how, and that less diverse teams typically have lower performance. The relative size of the giant component in a journal is associated with performance, with high-impact journals typically giving rise to larger giant components. However, in astronomy, neither p, q, nor S were significantly correlated with impact factor, indicating that this field is different from the others. The study concludes that team size evolves over time, possibly reaching an optimal size. Four of the five fields considered have similar values of p and q, suggesting the existence of a "universal" set of optimal values. The fact that astronomy has no correlations between p, q, or S and journal impact factor indicates that this field is different from the others. Whether these differences are due to the needs of the creative enterprise or historical or other reasons remains unclear.This study investigates how team assembly mechanisms influence the structure of collaboration networks and team performance in creative fields. The research focuses on both artistic and scientific domains, analyzing data from the Broadway musical industry (BMI) and four scientific disciplines: social psychology, economics, ecology, and astronomy. The study proposes a model for team self-assembly based on three parameters: team size (m), the probability of selecting incumbents (p), and the tendency of incumbents to repeat previous collaborations (q). The model predicts two phases in team assembly: one where agents form a large connected cluster and another where agents form isolated clusters. The results show that team assembly mechanisms determine both the structure of collaboration networks and team performance. The study finds that team size and composition depend on the complexity of the creative task. In the BMI, team size increased over time as the complexity of the musical form evolved. Similarly, in scientific collaborations, team size increased over time, with faster growth in ecology and astronomy. The analysis of team size cannot capture the fact that teams are embedded in a larger network, which acts as a storehouse for knowledge. The structure of the largest cluster in the network is influenced by the distribution of different types of links, with a higher proportion of repeat incumbent-incumbent links leading to a more homogeneous knowledge pool. The study also explores how team assembly mechanisms affect the performance of teams. It finds that successful teams have a higher fraction of incumbents, who contribute expertise and know-how, and that less diverse teams typically have lower performance. The relative size of the giant component in a journal is associated with performance, with high-impact journals typically giving rise to larger giant components. However, in astronomy, neither p, q, nor S were significantly correlated with impact factor, indicating that this field is different from the others. The study concludes that team size evolves over time, possibly reaching an optimal size. Four of the five fields considered have similar values of p and q, suggesting the existence of a "universal" set of optimal values. The fact that astronomy has no correlations between p, q, or S and journal impact factor indicates that this field is different from the others. Whether these differences are due to the needs of the creative enterprise or historical or other reasons remains unclear.