In the previous post, we presented a system for trading VXX, a volatility Exchange Traded Note. The trading system was built based on simple moving averages. In this post, we are going to examine the time series properties of VXX in more details. The figure below shows the VXX and its 200-day moving average for the last 5 years. From the Figure, it’s evident that the time series has a strong downward drift. The cause of this downward drift has been discussed to a great extent in the investment community. It will not be a subject of this post. Next, we calculate the daily returns of VXX and perform the Augmented Dickey-Fuller test to check the stationarity of the return series. According to Wikipedia,
We utilize Python to run ADF test on the return series. The test statistic is -19.5 and the p-value is zero. Therefore the return series is stationary. We then proceed to examine the autocorrelation functions of VXX returns.
The plot below shows the autocorrelation (ACF) and partial autocorrelation (PACF) functions of VXX returns. It’s interesting to note that VXX returns are positively auto-correlated at 3-day lag. It’s well-known that the stock indices are mean reverting in the short term. Here we observe the opposite, i.e. a trending property. Click on the link below to download the Python program and data file. Post Source Here: Stationarity and Autocorrelation Functions of VXX-Time Series Analysis in Python
0 Comments
## Leave a Reply. |