Underlying various routes to peer influence on risky choices is the assumption that individuals have beliefs around their peer's preferences which are incorporated in their choices. However, much is unknown about the accuracy of these beliefs and how they weigh in individuals' considerations. We tested these implicit assumptions by actually collecting real-life peers' preference to contrast with people's prediction, and quantifying what changes when individuals were asked to take the peer's perspective as the decision-maker instead of themselves. Since perspective taking develops through late adolescence, adolescence makes an especially dynamic window for observation. With a sample of typically developing friend dyads (N=128, 12.0-22.8 years), we collected fully mutual data on decision preferences in an economic risky decision making task with safe (certain) and risky (more variable outcomes) options that vary in their expected values. Upon establishing individuals' baseline risk preferences and their prediction of their peers' risk preferences, they took their own and their peers' perspective in choices where their unchosen option was assigned to the peer. We modified an economic expected utility model to include a new parameter representing the adjudication between one's own and friend's outcome, and analyzed age-related changes with Generalized Additive Models. We found although peer's risk preferences were overestimated in decisions on average, participants aged 16-22 years weighed friend outcome more and earned less when taking their friend's perspective compared to their own, indicating this is a heightened period for prosocial considerations.