Volatility & The Black Swan : Investigation of Univariate ARCH-models, HARRV and Implied Volatility in Nasdaq100 amid Covid19

Sammanfattning: Covid19 hit the world’s financial markets by surprise in March 2020 and ensuing volatility marked an end to the prior low-volatility environment. This Black Swan engendered numerous publications establishing how the equity market responded to the exogenous shock. However, there is no applicable comparison to Nasdaq100 regarding how models perform during extreme conditions such as ante, amid and post Covid19. Furthermore, goodness of fit together with forecasting accuracy are further examined in the light of new intra-day data from Oxford Man Institute covering this time-period. This thesis presents a comparison of volatility models incorporating economic intuition, sentiment, historical values of volatility and stochastics. By exploiting intra-day at 5 min interval the trade-off between noise and loss of valuable information effectively kept at a minimum yielding considerable robustness to the thesis’ result. Linear ARCH-models, Implied Volatility and HARRV applied with the addition of several different combinations of hold-out periods enable multiple vantagepoints for evaluation. This thesis finds HARRV’s series of one-step ahead prediction of future conditional volatility to be superior throughout all hold-out periods. I am able to present empirical evidence supporting the idea that HARRV’s additive cascades of volatility is superior to sentiment-driven implied volatility and ARCH-models pertaining to Nasdaq100.