RSM432H1 Lecture Notes - Lecture 3: Maximum Likelihood Estimation, Implied Volatility, Autoregressive Conditional Heteroskedasticity
Document Summary
Log return preferred over arithmetic mean return: normal distribution is foundation of many models but is not really representative of actual asset. Significant price shocks, jumps, and outliers make returns have excess kurtosis and be returns fat-tailed: volatility is time-varying, bad news is digested faster and has bigger impact than good news. There is a term structure of volatility. Leverage effect: volatility is measured by standard deviation of return per unit of time under continuous compounding. Usually quoted per annum and daily: need to consider historical volatility and ways of projecting future volatility, assume daily returns are independent with same variance, we can calculated implied volatility from the black-scholes model. This feeds into indices like the vix: vix communicates implied volatility to broader market. 20 is the historical threshold between volatile and calm markets. Vix calculated off of s&p futures options. Can also be used to calculate forward volatilities. Is market view today of future volatility for that asset.