AI Engineer NYC 2026
Join AI Engineer and Major League Hacking (MLH) on Sunday, October 11, for a full day of building, experimenting, and meeting fellow AI engineers.
TickerTales
AI agent that explains why your stock moved: a sourced, plain-English story for investors.
The problem
Stocks jump or drop every day, and beginners can’t tell why. The answer is scattered across dozens of articles, filings and posts.
What TickerTales does
- Stock story: enter a ticker and get a plain-English timeline of what moved the stock, matched to the price chart, with every claim linked to its source.
- Honest confidence: each report is labeled clear, likely or unclear. When no source explains a move, it says “no clear catalyst” instead of guessing.
- Chat: ask follow-ups like “explain this simply” or “dig deeper,” and it searches the web live.
- Personal watcher agents: follow your stocks, tell the agent what you care about, and get alerts personalized to you.
How it works
A LangGraph multi-agent pipeline: a price analyst finds unusual move days, Nimble searches news, trusted sources and social posts around those dates, three Claude Haiku analysts (news, earnings, sentiment) run in parallel, a Claude Sonnet lead writer builds the story, and a fact-checker verifies every citation.
Built with: Python, Streamlit, LangGraph, Claude (Sonnet + Haiku), Nimble Search API, Finnhub, yfinance.
For information and education only. Not investment advice.
Built with
AI tools
Prize Categories Entered
- Event prizeSubmitted
Best Overall
Beginners see stocks swing daily but can't find out why. TickerTales is a LangGraph multi-agent system (parallel Claude analysts, a lead writer, and a fact-checker) that turns live news, filings and social chatter into a sourced, day-by-day story, honestly labeled "no clear catalyst" when nothing explains a move. Users can chat with the report and create personal watcher agents that alert them based on what they care about.
- Event prizeSubmitted
Best Use of Nimble
Nimble is the research backbone, not a single search call. We find the exact days a stock moved unusually, then use Nimble's date-filtered news search to reconstruct what happened on each one. We also use trusted-domain search for credible sources, social search for retail sentiment, full-content retrieval so Claude reads whole articles instead of headlines, live "dig deeper" search in chat, and last-day news search for personal watcher agents. Every claim in a report links back to a page Nimble found.