Intro

GARCH modeling has been a topic of interest for many years, with much discussion around the fore- casting performance of the models in the context of equities, but relatively few papers have applied the technique to crytocurrencies. The rapid growth of Bitcoin over the past few years, coupled with its highly volatile price, makes it a perfect candidate for analysis. This paper produces daily forecasts of variance through an aggregation of intraday forecasts, and compares the performance against measures of realized variance. Aggregation methods are found to be more accurate than 1-step ahead daily forecasts

data

dis

Volatility model

\(r_{t_i} = log(P_{t_i}) - log(P_{t_{i-1}})\)

Descriptive plots of 5-minute and daily returns