Hi there:
About one year ago (from January 14th to January 25th), 2021… Blackberry (NYSE: $BB), Gamestop (NYSE: $GME) and Koss Corporation ( NASDAQ: $ KOSS), came to run very unusual (more than 300% in just a few days), clearly they were all economic bubbles… since they all fell in the days following this beheaviour.
Is there a time series to fit this kind of beheaviour in the markets??? or maybe the ARIMA model still works??
Particulary is there any time series that best fit the chart given in the picture of the following site:
https://www.listenmoneymatters.com/wp-content/community/uploads/2017/12/bubble2.jpg
This was clearly the pattern of those stocks about a year ago…
Hi Ghery, nice observation in #BB, #GME and #KOSS.
We don't have any time series that perfectly fit the price behavior you are referring to. However, we have several time-series forecasting models that can try to forecast the price by taking the price data as the input.
Out of all, the ARIMA model is observed to be the best time series prediction model we have, especially for short-term price prediction.
Please go through our Financial Time Series Analysis for Trading' course to gain a strong foundation on time series prediction models such as AR, MA, ARMA, ARIMA, and SARIMA.
You can try predicting the stock prices of #BB, #GME and #KOSS using the code given in our course.
If you want to explore on predicting the bubbles, you can refer to the references below.
In the literature, there are many attempts over the years to design a time series model that can signal stock price bubbles before they happen in real-time. The ideas include:
1. Using traditional unit test methods like Dickey-Fuller test and Augmented Dickey-Filler test in a repeated manner with rolling samples. Early stock bubble indicators were designed using these methods.
You can refer to the paper- Detecting asset price bubbles with time-series methods to know more about the study results.
2. This is an attempt to create a mathematical model for a 'bubble' using a stochastic differential equation and evaluating the occurrence of a bubble using an expectation function.
Refer to this paper to know more about this approach Method Of Forecasting Financial Bubbles.
3. Use the sup augmented Dickey-Fuller test (SADF) and the generalised sup augmented Dickey-Fuller test (GSADF) test to identify bubbles.
Here is an implementation of these tests. Testing Bubbles: Exuberance and collapse in the Shanghai A-share stock market
4. By understanding the factors behind the bubbles and collecting enough historical data across the exchanges over the years, we can try solving this problem using machine learning techniques.
Here is an attempt to predict stock market crashes using machine learning techniques-Predicting stock market crashes
Hope this helps!
Thank you
Thanks a lot… I really appreciate those links that you send me…