-
Full Graph Analysis
-
User Dashboard
-
Integrated Graph Analysis in Dashboard
Inspiration
- Financial news can be overwhelming, fragmented, and hard to act on — especially for students, beginners, or busy professionals. We wanted to create a platform that makes market insights clear, accessible, and actionable for anyone interested in trading or investing.
What it does
- MarketMentor aggregates financial news, clusters it into major events using AI, summarizes it in plain language, analyzes sentiment and potential impact, and predicts future events — all through a simple, interactive dashboard. Users can quickly understand what’s happening in the markets and get clear, actionable insights without needing prior finance knowledge.
How we built it
The backend is built with Python (Flask) and uses OpenAI models for summarization, topic extraction, and prediction. Financial news and stock data are pulled from external APIs like AlphaVantage and Polygon. News clustering is done using sentence embeddings. MongoDB stores persistent data, while diskcache speeds up API responses. Neo4j integration (in progress) will allow deeper relationship mapping between events, companies, and timelines
The frontend is built with React and TypeScript, featuring interactive dashboards that visualize news clusters, sentiment scores, predicted impacts, and advice. REST APIs connect the backend and frontend. We also built unit tests to ensure reliability across clustering, summarization, and prediction logic
Challenges we ran into
- Tuning the news clustering so that similar but slightly different articles were correctly grouped together
- Summarizing real-time news in a way that is both accurate and easy to understand
- Balancing fast API response times with heavy real-time data processing
- Designing a clean UI that could handle complex, interconnected data without overwhelming the user
Accomplishments that we're proud of
- Building an AI-powered platform that turns real financial news into clear, actionable insights
- Creating an intuitive visualization system to help users understand complex market movements at a glance
- Laying the groundwork for graph-based event relationship prediction with Neo4j
- Developing a project that lowers the barrier to entry for people curious or cautious about trading
What we learned
- How to integrate large language models into a real-time, user-facing application
- How to build effective clustering and summarization pipelines for noisy, real-world data
- The importance of clean UI/UX design in making complex financial concepts understandable
- The challenges of balancing performance and functionality when working with live APIs and AI models
What's next for MarketMentor
- Deeper personalization of insights based on users' portfolios and interests
- Integration with real trading platforms for actionable trading recommendations
- More advanced event prediction models using time-aware graph structures
- User feedback loops to improve predictions and advice
- Full authentication and authorization for user accounts
- Scalability improvements to handle larger datasets and higher user traffic
Log in or sign up for Devpost to join the conversation.