The economics of data: Using simple model-free volatility in a high-frequency world ⿿

This paper examines the practical implications of using high-frequency data in a fast and frugal manner. It recognises the continued widespread application of model free approaches within many trading and risk management functions. Our analysis of the relative characteristics of four model-free volatility estimates is framed around their relative long memory effects as measured by the feasible exact local Whittle estimator. For a cross-section of sixteen FTSE-100 stocks, for the period 1997⿿2007, we show that 5-min realized volatility exhibits a higher level of volatility persistence than approaches that use data in a sparse way (close-to-close volatility, high-low volatility and Yang & Zhang volatility). This observation is a useful decision-tool for a trading and risk management decisions that are undertaken in a time-constrained task environment. It recommends that the use of sparse data (open, high, low and closing price observations) requires trader intuition and judgement to build long-memory effects into their pricing.