Dashboard
Money Transfer
AI Invesment Hub
AI Trade History
Sentiment Intelligence
Market Data Agent

About TradrFin

Inspiration

TradrFin was born from a simple observation: financial news often says one thing while markets do another. We've all seen bullish headlines while prices plummet, or bearish news while markets rally. We asked ourselves: what if AI could validate sentiment against actual price action?

The idea was to build a system that doesn't just read the news. It cross-checks sentiment with technical indicators and flags contradictions. This led to our multi-agent architecture where specialized agents work together to produce more reliable insights.

We were also inspired by the hackathon tracks. We used Claude for intelligent analysis and Apify for scalable data collection. The challenge was combining them into a cohesive system that adds real value.

What it does

TradrFin is a multi-agent financial analysis system that validates market sentiment against price action. It uses four specialized AI agents:

News & Social Sentiment Agent: Collects financial news and analyzes sentiment using Claude
Asset Movement Diagnostics Agent: Computes technical indicators (RSI, MACD, ATR, Volume)
Cross-Verification Agent: Compares sentiment with price action and flags contradictions
User-facing Advisor Agent: Presents insights in plain English

The system detects when sentiment and price action disagree (e.g., bullish headlines but falling prices) and explains why. It provides:

Final market verdict (Bullish/Bearish/Neutral)
Confidence scores (0-1)
Plain English explanations
Actionable trading insights

Users interact through a React dashboard with real-time sentiment visualization, technical charts, and an AI chat interface.

How we built it

We built TradrFin in five phases:

Phase 1: Sentiment Collection (Apify Actor) Built a custom Apify Actor that collects financial news from Google News and Reuters feeds. Integrated Claude API to analyze sentiment, extract key themes, and generate sector-level insights.

Phase 2: Technical Analysis Engine Created the Asset Movement Diagnostics Agent using yfinance to fetch market data. Implemented indicator computation (RSI, MACD, moving averages, volume analysis) and built a precomputation pipeline for multiple symbols.

Phase 3: Cross-Verification Logic Designed the Cross-Verification Agent. Built logic that compares sentiment scores with technical indicators. When they conflict, the agent uses Claude to generate explanations. This is the "magic" that makes TradrFin unique.

Phase 4: User Interface Built a React dashboard with Tailwind CSS featuring:

Real-time sentiment visualization
Technical analysis charts
AI chat interface (Advisor Agent)
Responsive, modern design

Phase 5: Backend Integration Created two backend APIs:

Node.js/Express server (port 5000) for dashboard data and chat
Python/FastAPI server (port 8000) for trading simulations

Tech Stack:

Frontend: React, Vite, Tailwind CSS
Backend: Node.js/Express, Python/FastAPI
AI: Claude (Anthropic) for sentiment analysis
Data: Apify Actor, yfinance, Finnhub
Database: TiDB
Development Environment: Cursor

Challenges we ran into

1. Data Quality & API Reliability Financial APIs can be unreliable. Endpoints fail, rate limits block requests, and data formats vary. We solved this by implementing fallback mechanisms, caching strategies, and using multiple data sources (Finnhub, yfinance, AlphaVantage) as backups.

2. Real-time vs. Historical Data Synchronization Balancing real-time sentiment with historical price data was tricky. We precomputed indicators for historical analysis while keeping sentiment collection real-time, requiring careful synchronization logic.

3. Cross-Verification Logic Design Designing rules that detect contradictions without false positives was challenging. We iterated on the logic, tested numerous edge cases, and refined Claude prompts to generate accurate, human-readable explanations.

4. Multi-Agent Coordination Getting four agents to work together smoothly required clear interfaces and robust error handling. We learned to design resilient systems where one agent's failure doesn't break the entire pipeline.

5. Time Constraints Building a full-stack multi-agent system in a hackathon timeline was intense. We prioritized core functionality, used existing libraries strategically, and focused on making the cross-verification feature shine.

6. Environment & Dependency Management Managing different environments (Node.js, Python, Apify) and their dependencies was complex. We documented setup steps carefully and created clear separation between components.

Accomplishments that we're proud of

1. The Cross-Verification Agent The core innovation that validates sentiment against price action. It detects contradictions and explains them. This is what makes TradrFin unique.

2. Working Multi-Agent System Successfully orchestrated four specialized agents working together. Each agent has a clear role, and their collaboration produces insights no single agent could generate alone.

3. Apify Actor Integration Built and deployed a custom Apify Actor for scalable news collection. This was our first Apify project, and it works seamlessly with the rest of the system.

4. Claude Integration Effectively leveraged Claude for both sentiment analysis and intelligent cross-verification. The prompts generate structured insights and human-readable explanations.

5. Full-Stack Implementation Delivered a complete system from data collection (Apify) to AI analysis (Claude) to user interface (React dashboard). Everything works together end-to-end.

6. Judge-Friendly Output The system produces clear, actionable insights that judges can understand. Example: "Sentiment is bullish but price action contradicts it. The market may be in distribution phase."

What we learned

Multi-Agent Architecture: Designing systems where specialized agents collaborate requires clear interfaces, error handling, and data contracts. Each agent should be independent yet work together seamlessly.

Claude API Mastery: Learned to craft effective prompts that extract structured insights from unstructured data. Claude's reasoning capabilities are powerful for cross-verification logic.

Apify Platform: Built our first Apify Actor and learned the actor model for scalable data collection. The platform makes it easy to deploy and run data collection independently.

Technical Analysis: Implemented financial indicators (RSI, MACD, ATR) and learned their interpretation. Balancing simplicity with accuracy was key.

System Design: Learned to build resilient systems with fallbacks, caching, and multiple data sources. One API failure shouldn't break the entire system.

Time Management: Prioritized core features and made strategic trade-offs. The cross-verification feature was worth the extra effort.

Full-Stack Integration: Connected frontend, backend, AI services, and data collection into a cohesive system. Clear API contracts and error handling were essential.

What's next for TradrFin

Short-term improvements:

Real-time streaming data integration for live market updates
More technical indicators (Bollinger Bands, Stochastic Oscillator)
Historical backtesting to validate cross-verification accuracy
User authentication and portfolio tracking

Advanced features:

Machine learning models to improve contradiction detection
Social media sentiment analysis (Google News, Reuters)
Automated trading alerts based on cross-verification signals
Portfolio optimization recommendations

Platform enhancements:

Deploy Apify Actor to Apify Store for public use
API rate limiting and caching optimization
WebSocket support for real-time updates
Advanced visualization dashboards
Export functionality for trading logs

Research & development:

Study correlation between sentiment-price contradictions and market outcomes
Refine confidence scoring algorithms
Build a community around the platform
Open-source components for the developer community

Long-term vision: TradrFin could become a platform that helps traders make better decisions by validating sentiment against reality. We envision a community-driven system where users contribute insights and the AI learns from market patterns.

Built with passion, curiosity, and a lot of coffee ☕