Volatility Forecasting in Bull and Bear Markets: Evidence from the US stock marketTools Sideris, Epameinondas (2016) Volatility Forecasting in Bull and Bear Markets: Evidence from the US stock market. [Dissertation (University of Nottingham only)]
AbstractThis thesis considers the performance of variance forecasting in US market index during bull and bear markets. The market index I investigate is the Standard & Poor’s 500 and the bear period I examine is from 15/8/2000 until 30/1/2003 and the bull period is from 3/3/2003 until 30/3/2007. The techniques I employ to make the forecasts are a) an Exponential Weighted Moving Average, b) implied volatilities from the official volatility index VIX, c) a GARCH (1, 1) and d) a EGARCH(1, 1). Both GARCH models are applied in t –student and general error distributions. The forecasting horizon is one day ahead for both market states using daily data and the realized volatility is approximated by the Parkinson model. Performance for each model is measured by the MSE, MAE and HMAE loss functions. I find that EGARCH model performs best in both bull and bear markets according to MSE and HMAE loss functions, while according to MAE, the EWMA is the best performing model in bull market. On the other hand, in both market states implied volatility make the worst predictions. In general, the forecasts are more accurate during the bull market compared to bear market.
Actions (Archive Staff Only)
|
Tools
Tools
![[img]](/style/images/fileicons/application_pdf.png)
