We measure financial literacy in a large sample of LinkedIn members, complementing a standard set of questions with a method that allow us to isolate and distinguish optimism and self-confidence. Like previous work, we find that high literacy respondents are more likely to save for a rainy day, plan for retirement, and are more likely to pay attention to fees when choosing credit cards. However, this is mostly driven by perceived, rather than actual, financial literacy: controlling for self-perceptions, actual literacy has low power to predict financial engagement. Moreover, behavior biases drive participation among low literacy respondents and are associated with mistaken beliefs about financial products and a lower willingness to accept financial advice. This has important implications for policy and for the design of institutions aimed at increasing literacy and protecting consumers from fraud.