摘要： This study builds a series of models to predict trading volume in European markets using different statistical methods. The analysis considers a number of aspects, such as special events (e.g. MSCI rebalances, futures expiries, or cross-market holidays), day-of-the-week effects, and the volume-price relation asymmetry, in order to perform contextual one-step ahead prediction. We investigate the prediction error for each calendar circumstance to infer a cross-stock event-oriented switching model for volume prediction. The study concludes by proposing a stock-specific out-of-sample metamodel that is fit by selecting an initial stock-specific model yielding the best performance for the most recent observations.
关键词：Trading volume; expiry day effect; holiday effect; behavioral finance; European stock market; feature selection
