We study the degree of individual and aggregate market overreaction in a dynamic experimental auction market. In 13 sessions with overall 101 students we find overreaction to new information both in stock price forecasts and transaction prices. Interestingly, market forces do not seem to help in lowering overreaction to new information in our setting. Moreover, we illustrate that subjects are not able to learn from their previous failures and thus do not correct their erroneous beliefs. Hence, overreaction in our setting remains on a stable level although subjects can at least in theory learn from other market participants or from outcome feedback. Lastly, we find first experimental evidence for a positive relation between differences of opinion and trading volume in a continuous auction market with several market participants.