EN :
Traditional techniques in portfolio management rely on the precise knowledge of the underlying probability distributions; in practice, however, such information is difficult to obtain because multiple factors affect stock prices on a daily basis and unexpected events might affect the price dynamics. To address this issue, we propose an approach to dynamic portfolio management based on the sequential update of stock price forecasts in a robust optimization setting, where the updating process is driven by the historical observations. Forecasts are updated using only the most recent data when the stock price differs significantly from predictions. In this work, we present a robust framework to optimal selling time theory. We introduce a wait-to-decide period, and allow actual price movements to drive the best decision in response to a bad investment. Numerical results illustrate our strategy, which requires less frequent updating of the problem parameters than in the traditional approach while exhibiting promising performance.