-
Top Stock Picks
-
Text Analysis
Inspiration
Was inspired due to the recent events of a lot retail investors joining and the need to understand this type of data in the stock discussion communities
What it does
Given a subreddit, MarketBuddy scrapes hot posts off there, filters out the top stocks, and analyzes post texts to state whether a subreddit is bullish or bearish.
How we built it
The frontend was built in React using Semantic UI while the backend was just a Flask API connected with the Reddit Python library to help get posts and comments.
Challenges we ran into
Building a valid model to classify text as bullish or bearish, there were lots of complications in finding out the intent and training the model.
Using random forests with attribute selection, I wasn't fully able to implement the NLP model as it needed more time to be trained properly on labeled data
Accomplishments that we're proud of
Having a working product across three different subreddits
What we learned
Learned a whole lot about full-stack dev and the new Flask framework!
What's next for MarketBuddy
Build out a proper text-based model to classify post submissions
