High frequency volatility variation as a measure of performance for ETFs

Abstract

In this paper, I analyze the US ETF market on high-frequency intraday volatility by dissecting positive and negative volatility. Previous research shows that by measuring volatility at a high-frequency rate it is possible to measure short-term jumps which can be described as ‘good’ or ‘bad’ volatility (Bollerslev et al. (2020)) based on its previous returns. The difference between positive and negative volatility, defined as realized signed jump (RSJ) follows the characteristics of skewness where previous research shows how skewness can be a predictor of future returns (Amaya et al. (2016)). In essence, ETFs are diversified portfolios where it is well documented that such have a negative skewness. Combined with investor's preference for positive skewness, this paper shows that ETFs can show strong performances using skewness measures. The difference between positive and negative volatility and its higher moments are used to calculate realized measures which are then used to measure the performance of different portfolios. Sorted quintile portfolios are used to measure the performance differences between the highest and lowest quintiles and show its statistical significance. Robustness checks are performed to show that by adjusting for large cap ETFs and sorting by asset classes further significant performance can be achieved for skewness measures.