CapitalBytes

• 

  Sample of a Positive Sentiment

• 

  logo

• 

  Chat with a robo advisor!

Inspiration

We wanted to simplify the experience of judging stocks quickly and in a user-friendly way!

What it does

The most important thing the project does is it provides financial insights in a quick way. The application showcases summarized news articles, gives stock forecasts, and gives sentiment analysis on news articles.

How we built it

The application uses Google Cloud Natural Language and python's nltk to summarize articles, we used the GIPHY API for the gifs hosted on the application. The application itself was built using Flutter. We used a MongoDB Atlas backend to store company and stock information. We used Keras and Tensorflow to train a recurrent neural network to forecast stock prices on the data from the AlphaVantage API.

Challenges we ran into

Some of the biggest issues we faces involved MongoDB Atlas and using Google Cloud Compute.

Accomplishments that we're proud of

We've never used MongoDB, so we were happy to have included that into the project, and the fact that we were able to predict a stock price using recurrent neural network was a first as well.

What we learned

We all learned something different, the biggest things we learned involved Google Cloud Compute, learning to generate summaries, creating and training recurrent neural networks and setting up MongoDB.

What's next for CapitalBytes

Get more data to train the stock forecaster over more iterations, iterate on the text summarizer, and work to have a more elegant GUI.

Built With

• deep-learning
• flutter
• google-cloud
• heroku
• machine-learning
• mongodb
• nltk
• python
• react
• scikit-learn
• spacy