Learning Fast or Slow

We analyze the performance of and learning by individual investors who engage in day trading in Taiwan from 1992 to 2006 and test the proposition that individual investors rationally speculate as day traders in order to learn whether they possess the superior trading ability. Consistent with models of both rational and biased learning, we document that unprofitable day traders are more likely to quit and that day traders begin with relatively small trades that increase as they gain experience. Inconsistent with models of rational speculation and learning, we document that the aggregate performance of day traders is negative and that over half of day trading can be traced to traders with considerable experience and a history of losses.