Hello folks! In this video we will talk about Quantra’s online course on Financial Time Series Analysis. Consider a situation where you meet a friend after a year or so. They might comment that you have changed a lot. But for you, not much has changed. You still feel the same. Over the days, the change in you has been so gradual that you don’t realise any difference. Let us take another example. You want to read a book on Financial Time Series Analysis and complete it in a month. It is a 600 page long book. The idea of finishing it in a month is daunting. But then, you apply a trick of learning. You promise yourself to read 20 pages every day, at a fixed time slot. And you do it.
You read the entire book in a month. Next month, when you promise yourself to read another book of 1000 pages, you are confident. Using your past behaviour, you make a fair estimation of the future. Time has an inherent nature to store information. A time series is simply to have data stored at regular time stamps. It could be weather records. It could be patient health monitoring, such as in an electrocardiogram (ECG). Or it could be financial time series such prices of a stock. Let’s look at an example of a stock now. Apple was trading at $24 in January 2016. But it is trading at $132 in January 2021. That’s more than five times the price in 2016! If someone looks at these two figures, they will say it is impossible to predict that Apple could reach $132.
They are right in this regard. However, if you look at the daily price data of Apple over the years, you will find it is not exactly impossible. You can try to find some pattern in this graph and accordingly forecast the future price. This is the basic principle behind time series analysis. In order to perform time series analysis, you need to make sure that the provided data is periodical in nature. If we are looking at the closing price of Apple for the past 5 years, we can’t have the year 2018 missing. Also, it cannot be a mixture of close and open prices. In that case, we will not get a good estimate.
After receiving this data, you should try to find any patterns in the data. Broadly speaking there are four components or patterns of a time series Seasonal, Cyclical, Trending and White Noise. Identifying these patterns will help you choose the correct time series model for analysis. Before you start time series analysis, you should make sure that the past values do have some impact on the present values. How will you check that? You use the Auto Correlation Function (ACF) and Partial Auto Correlation Function (PACF) plot for this purpose. Depending on the results, you will choose a time series model to use for forecasting future prices.
You will learn different time series models in an intuitive manner and also their strengths and limitations. The model names are shown on screen. Now, these models focus on the price series as the parameter to forecast. Is there another parameter which can be used for forecasting in time series analysis? Yes, volatility can also be used. You know that Tesla Inc. has large price changes on a day to day basis. Compared to Coca- Cola which is stable, Tesla’s price can increase or decrease as much as 15%. This is why Tesla is said to have high volatility. You can use time series to analyse the Volatility Index (VIX), created by the Chicago Board Options Exchange (CBOE).
You will use models like ARCH and GARCH for this purpose. This will give you an edge when you are trading VXX. Volatility forecasting models also play a role in other aspects which are shown on screen. In this manner, you can perform time series analysis on different data sets and create your own unique trading strategy. What’s more! You can also use the strategies to live trade as Quantra courses have in-built support for live trading. What are you waiting for? Go Quantra!.