Daily Market Commentary Agent
One button turns live market data into a finished daily client email
What it is
An agentic Next.js app for a financial-advisory firm that replaces a manual morning routine of uploading files to a Claude Project chat. One button pulls 14 live data sources, computes the technical analysis in TypeScript, renders a SPY chart, and has Claude write the daily market commentary as plain-text and styled HTML emails archived in Supabase. It also maintains the support and resistance ladder, swing-point history and a rotating analogy category so each day answers the previous day's cliffhanger.
The problem
A market strategist spent each morning manually assembling data and files for an AI chat to produce the daily client commentary.
How it works
- The TypeScript layer computes all analysis math (pivots, support and resistance zones, regime-change checks); the Python chart route only draws it and Claude vision confirms the reading but never derives it.
- Data is fetched in parallel from FMP, FRED, Yahoo and CNN Fear & Greed, and market breadth is computed in-app from S&P 500 constituents.
- Supabase Postgres tables (pivots, support_resistance, sr_change_log, emails, runs, settings) and Storage buckets hold state, email archives and charts.
- The firm's master instructions are embedded as the system prompt, and each run reads the prior email so the new one closes pending watch levels.
Stack
Disciplines
For your business
A one-click daily research and client-communication pipeline that gathers data, runs the analysis and drafts the finished email for human review.