Efficient Portfolio Allocation with Sparse Volatility Estimation for High-Frequency Financial Data

Traditionally, investors try to estimate short term portfolio volatility based on daily return. When tick-by-tick data are available, investors use different volatility estimators based on high-frequency data to evaluate the portfolio risk in the hope of outperforming those based on low-frequency data. In this paper, we optimize block realized kernel estimator in Hautsch et al. (2015) and propose another more efficient way when we deal with the large portfolio allocation. Our research contribution focuses on the benefits of high-frequency data for portfolio allocation based on sparse volatility estimate methods. This process provides us new insights and alternatives when we want to set up a sensible investment strategy especially for risk averse investors.