work in progress- please do not quote without permission To study gender di¤erences in overcon…dence, we use a two-stage decision model. This model allows for the coexistence of an overcon…dence bias and a strategicallydriven bias in the forecasting process. Our results con…rm the gender stereotype that male analysts are overcon…dent, while female analysts are not. Male …nancial analysts overweigh the precision of their private information by about 3%, believing too much in their own abilities. Female analysts, on the other hand, do not attach a weight to their personal information that is too high compared to the rational Bayesian weight. Moreover, both male and female analysts strategically in‡ate their earnings forecasts, but this strategic behavior disappears after the implementation of the new analyst regulation 2002.