New forecasting method reveals surprising disconnect between implied and historical volatilities.
The article explores different ways to predict how stock prices will change by looking at how volatility behaves. They use a method to estimate the unknown volatility function based on noisy observations. The quality of predictions depends on the type of volatility function - if it often fluctuates, one method works better, while another is better for smoother functions. When comparing implied volatility with historical volatility for five companies, they found no statistical relationship between the two.