Signal

Investing isn’t just numbers—it’s narratives. We zoom out across fundamentals, news, sentiment, and bias to deliver the full market story.

Comment

Inspiration

The idea for Signal came from a real investing dilemma.

At a hedge fund, we saw a company with extraordinary financial potential. On paper, the fundamentals were strong. The growth trajectory was compelling. But there was one variable traditional models couldn’t quantify: public perception.

The CEO’s media presence and political visibility significantly influenced how the market viewed the company. Despite strong financials, reputational risk changed the investment decision.

That moment raised a question:

What if an AI agent could analyze not just financial fundamentals, but also public narrative, media bias, and sentiment — the way real investors actually think?

Signal was built to bridge that gap.

What it does

Signal is an AI-powered investment decision agent.

Instead of predicting stock prices, Signal helps users decide whether a company is:

  • ✅ Invest
  • ⚠️ Risky
  • ❌ Avoid

It combines:

  • Financial fundamentals
  • Real-time news
  • Sentiment analysis
  • Political/media bias detection

And outputs:

  • A clear recommendation
  • Structured reasoning
  • Key risk factors
  • Supporting headlines
  • A chatbot for follow-up questions

Signal doesn’t just analyze numbers — it analyzes narrative.

How we built it

Frontend: Next.js + TypeScript + Tailwind

Backend: FastAPI (Python)

Database: MongoDB

AI Layer:

  • K2 Think V2 → structured reasoning + decision synthesis
  • Hermes (Nous) → conversational explanations

Pipeline:

1. User enters a company.

2. Backend fetches:

  • Financial summary (via API)
  • Recent news articles

3. NLP layer:

  • Sentiment scoring
  • Bias detection

4. AI reasoning engine synthesizes:

  • Financial signals
  • Narrative signals
  • Risk profile

5. Output: recommendation + explainable reasoning.

We designed the system to mimic how analysts combine quantitative and qualitative signals.

Challenges we ran into

1. API Integration & Live Data

  • Coordinating financial APIs and news APIs in real time
  • Handling inconsistent data formats
  • Managing rate limits

2. Quantifying Narrative Risk

  • Sentiment alone isn’t enough — we had to differentiate between:
  • Short-term noise
  • Structural reputational risk

3. Avoiding Overengineering

In a 24-hour build, the biggest challenge was deciding what not to build. We prioritized clarity, explainability, and demo impact over complexity.

Accomplishments that we're proud of

  • Building a fully functioning AI decision agent in 24 hours
  • Creating a system that feels aligned with how real investors think
  • Designing an interface that makes complex signals intuitive
  • Turning financial + narrative data into actionable insight

Most importantly, we built something meaningful to us — at the intersection of fintech, AI agents, and decision automation.

What we learned

  • Markets are driven as much by perception as performance.
  • AI agents are powerful when they synthesize multi-modal signals.
  • Simplicity wins in hackathons.
  • Explainability builds trust.

We also learned how to rapidly integrate APIs, structure AI reasoning prompts, and design for demo impact under tight time constraints.

What's next for Signal

  • Add historical narrative-impact tracking
  • Integrate prediction market signals
  • Build a portfolio-level risk dashboard
  • Add “What changed?” alerts when sentiment shifts dramatically
  • Deploy as a browser extension for retail investors

Long-term, we envision Signal as a personal AI investment co-pilot — one that understands not just the balance sheet, but the story behind it.

Built With

Share this project:

Updates