Project uses ML techniques to predict future values of stocks. In order to do this, it involves LSTM Network Architecture and data from Yahoo Finance API.
The following software requires following modules to be installed. Please see requirements.txt, or install all modules using the command:
pip install -r requirements.txt
To use the software, run the following command:
python main.py <stock-ticker-symbol>
| Company | Ticker Symbol |
|---|---|
| Microsoft Corporation | MSFT |
| Intel Corporation | INTC |
| Apple | AAPL |
| General Motors | GN |
| Alphabet (GOOGLE) | GOOG |
| NVIDIA Corporation | NVDA |
This stock prediction app analyzes the previous changes in the company's stock value. The model is trained on data reaching deep into past 10 years.
Model surprisingly manages to predict the shape of future stock plot, but it does not apply to every company.
There are several factors that influence the deceitfulness of this prediction app. One of them is the changing popularity of the company.
See for example TESLA, which is most definitely a successful business whose main founder (Elon Musk) is a very popular celebrity on the social media. His latest posts (tweets) can affect the stock price of "his" company.
Here is the validation of the predicted model for the period of 2020.01 - 2021.07. Black lines indicates the real past values, green are the predicted ones for that period.
It is clearly visible that the model doesn't include some important factor(s) that lead to the false predictions. The event that is probably shown on the plot might have been introduction of the new tesla car or some other technology related to car's autonomy. Of course there are many more reasons why this stock validation was false (e.g. company innovations).
On the other hand, there are companies that experience increase in some repeatable periods. E.g Apple as a big tech company releases new products mostly at the same time intervals (e.g. )
| Models | Release date |
|---|---|
| iPhone 12 series | October 13, 2020 |
| iPhone 11 series | September 10, 2019 |
| iPhone XS series | September 12, 2018 |
| iPhone 8 series | September 12, 2017 |
It might be the reason why the predicted validation of the past year's data came out well enough.
Another common factor that leads this model to false predictions, is the lack of the ability to predict future
demand. Without a doubt we can see that the covid-19 pandemic affected many sectors of technology industry. One of the effects of the pandemic was an increased demand on technology hardware that caused Global Chip Shortage of 2020-21.
Companies whose profits are based on production of such technologies experienced increase of its stocks value.
Events like this might introduce some misstatements into the prediction model.
Image shows General Motors stocks predition 30 days into the future.
