MarketMood

Inspiration

As NLP becomes an evermore powerful tool, we wished to apply its strengths to an age-old problem of predicting the market.

What it does

MarketMood is composed of 3 main components. A web-scraping server that automatically gathers the top news articles on a by-stock basis every 10 mins and uses Cohere's NLP API to determine investor sentiment, a backend that stores relevant statistics and predictors, and a user-friendly frontend that displays the data in a visually appealing manner.

How we built it

We built this project in many different technologies, including BeautifulSoup for web-scraping, the Cohere API for NLP, Flask to host the server for consistent web-scraping, PyMongo to store data, and ReactJS + Tailwind for the frontend.

Challenges we ran into

We rethought our entire product to create MarketMood within the last 12 hours of this hackathon, for one. Moreover, many of our team members used technologies for the first time, including Cohere, MongoDB, React, and more.

Accomplishments that we're proud of

Building a fully functional product!

What we learned

Many different technologies, that it's better to start your project earlier than later, and coffee does work.

What's next for MarketMood

MarketMood can be extended and built upon in many different ways - in particular we can begin using collected data from the past to improve our future predictions, create a user registration feature to allow instant updates about the market, and implementing continuous testing to enhance the performance of our market sentiment algorithms.

Built With

• beautiful-soup
• cohere
• flask
• mongodb
• natural-language-processing
• python
• react
• tailwindcss