GrowMore

What it does

The web application recommends a user whether to buy, hold, or sell a stock based on specific indicators. Based on the tweets from the CEO and the company page it calculates the polarity of the tweet. It uses this polarity with the historical data to predict the price of the stock tomorrow. Finally, based on the indicators used, users are provided recommendations.

How we built it

Our process started with gathering data from Yahoo Finance and Twitter API. We performed cleaning and analysis of the data and integrated the historical stock data with the polarity of the tweets. Performed feature engineering and hyper-parameter tuning for the Long Short Term Memory model. Validated the model based on root mean squared error metric. Based on the stock price prediction of the next day and indicators such as ... we provide recommendation to users. The app is built using Streamlit, the model is built using Tensorflow.

Challenges we ran into

The major issue that we faced during the projects was what parameters to consider while suggesting whether to buy or sell a stock.

Accomplishments that we're proud of

Integrating sentiments with the historical stocks data.

What we learned

We learned how to work with Time series data and integrate it with sentiments. Moreover, we also learned about various parameters such as Moving average, Standard deviation, RMSE based on which one can recommend whether to buy, sell or hold stock.

What's next for GrowMore

Allow the website to invest a specific amount for each stock based on the recommendation.
Automate the model to fetch daily data and train accordingly.
Work on improving the recommendations.