MacroMind

Inspiration

As a high school student passionate about finance and technology, I was fascinated by the amount of data available to retail traders but frustrated by the lack of tools to make sense of it all. I saw traders on platforms like Reddit drowning in news headlines, conflicting opinions, and complex charts. The core problem wasn't a lack of information; it was a lack of clarity. I was inspired to build MacroMind to solve this problem, creating an "insight engine" that could automatically connect the dots between market sentiment, economic events, and price action, providing the kind of analytical edge previously only available to large financial institutions.

What it does

MacroMind is an AI-powered financial analytics platform designed to give traders a clear data-driven edge. It analyzes the "what and why," by ingesting real-time news and social media data, using FinBERT and a GPT LLM to not only score market sentiment but to distill the key themes and tell the user why sentiment is changing. It forecasts "what's next" by using a machine learning model (XGBoost) to analyze market data and sentiment patterns to generate a multi-faceted forecast of future volatility, including its level, trend, and the overall market regime. It empowers users to act on these combined insights and answer, "so what?" by setting highly specific, multi-conditional alerts (e.g., "notify me only if volatility is high AND sentiment turns negative").

How I built it

• Backend: A fully asynchronous API built with Python and FastAPI, designed for high performance.
• Database: PostgreSQL for robust and reliable data storage, managed with SQLAlchemy and Alembic for migrations.
• AI & Machine Learning:
• Hugging Face Transformers (FinBERT) and LangChain/OpenAI for the "Insight Distiller"
• A custom XGBoost regression model built with Scikit-learn and Pandas for feature engineering
• Facebook Prophet for pattern recognition features
• Frontend: Next.js framework with Typescript and CSS styling

Challenges I ran into

As a solo founder, the biggest challenge was prioritization. My vision for MacroMind is huge, but I had to focus on a core, high-impact MVP for this hackathon. Deciding which features to build now versus which to defer was a constant strategic challenge. The most complex task was the data engineering, which was creating a reliable pipeline to ingest, clean, and align time-series data from disparate sources (market prices, news timestamps, sentiment scores) to feed the machine learning models accurately.

Accomplishments that I'm proud of

I am most proud of building this quite ambitious project on my own, but specifically, creating the end-to-end volatility forecasting service, from feature engineering in Pandas to training and serving an XGBoost model, was a major accomplishment for me and went beyond simply calling an AI API. Designing a product that uses an LLM to provide qualitative explanations for quantitative sentiment scores is a feature I believe provides a lot of value and really differentiates MacroMind.

What I learned

This project has been an incredible learning experience in software engineering, finance, and business strategy. I learned that building a successful AI product is 10% model training and 90% data engineering and MLOps. Creating robust data pipelines is just as important as the model itself. I also learned the critical importance of validating a problem before building a solution. For this hackathon specifically, I learned how to distill a large, complex project into a short, compelling pitch and demo, which is a crucial skill for any founder or entrepreneur.

What's next for MacroMind

1. Allow for the access of all the rest of the implemented API endpoints through the frontend (like economic calendar, opportunities, etc.)
2. Enhance the frontend and add more customizability
3. Launch a closed beta and onboard the first 100 beta users from target communities to gather feedback and validate the core value proposition
4. Implement the subscription pricing model
5. Evolve the AI and expand its capabilities, as well as allowing it access to more sources and higher-quality data as revenue increases
6. Fully launch the content and community engagement strategy to build a loyal user base and drive organic growth

Built With

• fastapi
• finbert
• javascript
• langchain
• newsapi
• next.js
• openai
• postgresql
• prophet
• python
• react
• redditapi
• redis
• typescript
• websocket
• xgboost