The project Idea comes from the need for AI-based solutions in Stock Market. It is very crucial to use the past data with the current market sentiments to find the correct predictions for the stock prices of the company. We wanted to design a solution by leveraging past sentiment values extracted from Twitter posts and historical stock data available at different open source platforms.

What it does

It is a web application that helps a user to manage the stock market portfolio. Users can buy/sell stocks, See the trends in the market, can watch the history of stocks in various formats, and get predictions whether they should buy stocks or not. Our solution uses a Machine Learning model trained using Long Short Term Memory (LSTM) and Recurrent Neural Network(RNN).

How we built it

We used Django Framework to build the backend of the system. React JS is used to visualize the data and manage other front-end activities. We used LSTM with RNN to train datasets generated using sentiments and historical stock data to create a prediction model for stock prices.

Challenges we ran into

Data Generation for sentiments was one of the difficult challenges since extracted data was limited to 3200 posts by Twitter API. we had to start again using another API called Twint. Also, we had to train the model on the dataset in very little time.

Accomplishments that we're proud of

We were able to achieve all the milestones that we planned to deliver in the beginning. We have developed a machine learning model, a front-end web application, and a back-end web application in 24 hours. That is really proud thing to do.

What we learned

We have performed the task in a team of 4. We have learned about working with tight deadlines. We have also learned how we need to perform operations related to machine learning in a short period of time.

What's next for Atrade Stock Exchange

We have huge plans for this system. This system can be extended to a proper prediction platform using Better Machine Learning Model that includes time series and a bit longer memory storage while training. Also, the web application can be extended to the mobile application. We can add many features as recurring buy, risk management, sentiment, and historical data control, etc.

Share this project: