In principle, as the frequency of the data grows larger, the quality of volatility forecasts should improve. In this paper we propose to use three indicators of volatility and to analyze the dynamic interactions between them. We compare the outcomes obtained with two different model selection procedures and we show the performances of the models in terms of volatility forecasting over a month horizon by resorting to a market-based volatility measure such as VIX. The results show that the variables derived from the multiple indicators offer explanatory power both in and out of sample.