I design a laboratory experiment to study why experience fails to prevent retail investors from trading speculatively and suffering losses in expectation.  Subjects in the experiment observe private information and then decide whether to swap Arrow securities with a partner. A no-trade theorem applies to the setting so that under rational expectations, trade should never be realized. I show that experience reduces trade resulting from overconfidence, but fails to correct strategic naivete, as experienced subjects continue to ignore the selection bias implied by their partners’ willingness to trade. This result is most salient for subjects with high-quality information, who trade more frequently after learning their information reliably predicts the state of the world but do not choose to trade less when learning their partners’ information is similarly high-quality. After revealing their partners’ information about the state of the world—thus removing the role of strategic naivete—I find subjects are less willing to trade and are more responsive to their partners’ information quality. My results imply retail investors lose from trading because they fail to consider the information driving others’ trading decisions and repeated experience does not fully correct this bias.