What It Does
We utilize regression on four metrics to predict financial earnings based on quarterly historical data. With an LLM for sentiment analysis, we predict financial performance on a scale of positive, neutral, or negative.
Technologies We Used
We used next.js, mantine, typescript, react chartjs, cohere, google custom search, transformer, beautiful soup, other Python packages, Reddit API, yfinance API, and packages such as requests, flask, pandas, etc.
Challenges we ran into
The most prominent challenge was identifying data. We found that most news source APIs require payment, and any free tier either requires manual verification for educational use or is limited in the ability to backtrack beyond a month. To address these challenges, we identified alternative resources like social media sources and google custom searches, which provided enough data to support analytics over a sample stock.
Built With
- beautiful-soup
- cohere
- flask
- github
- google-custom-search
- mantine
- nextjs
- requests
- taillwindcss
- transformer
- typescript
- yfinance
Log in or sign up for Devpost to join the conversation.