[Statistical model selection criteria provide an informed choice of the model with best external (i.e., out-of-sample) validity. Therefore they guard against overfitting ("data snooping"). We implement several model selection criteria in order to verify recent evidence of predictability in excess stock returns and to determine which variables are valuable predictors. We confirm the presence of in-sample predictability in an international stock market dataset, but discover that even the best prediction models have no out-of-sample forecasting power. The failure to detect out-of-sample predictability is not due to lack of power.]

Before the opening of the Paris Bourse, traders place orders and indicative prices are set. This offers a laboratory to study empirically the tâtonnement process through which markets discover equilibrium prices. Since preopening orders can be revised or canceled before the opening, indicative prices could be noise. We test this against the hypothesis that preopening prices reflect learning.Early in the preopening the noise hypothesis is not rejected. As the opening gets closer, the informational content and efficiency of prices increase and the learning hypothesis is not rejected. We also propose a GMM‐based estimate of the speed of learning.

Statistical model selection criteria provide an informed choice of the model with best external (i.e., out-of-sample) validity. Therefore they guard against overfitting (“data snooping”). We implement several model selection criteria in order to verify recent evidence of predictability in excess stock returns and to determine which variables are valuable predictors. We confirm the presence of in-sample predictability in an international stock market dataset, but discover that even the best prediction models have no out-of-sample forecasting power. The failure to detect out-of-sample predictability is not due to lack of power.