 |
The analysis of volatility is paramount for decision-making in financial markets. Here is developed a strategy for increasing information to be used in the build of relevant volatility measures. With the increased information available through the new technologies most of the data analyses in finance are ``big data'' problems. The time deformed returns associated with the volume are used to estimate and forecast intraday high-frequency volatility. This kind of return can give a new perspective on volatility evolution, because allow estimation, forecasts and decisions to be considered at a varying speed when measured in calendar time, which is compatible with the reality in financial markets, periods of high and low activity are clearly identifiable. Through this strategy, other information elements can be extracted from the data, and not only the traditional fixed time-interval prices usually obtained. By extracting other information related to volatility evolution, for example, volume of trade and durations, better volatility estimates can be constructed. It allows the generalisation of models, namely the ones that include the possibility of jumps in the volatility, which without such increase of sample information present characteristics of underidentification, leading to inconsistent estimates and creating biased forecasts.
