finny is a financial analysis notebook built on Claude Code. You work from your terminal or VSCode.
Use finny to run trading comps, benchmark companies, review Excel models, gather news from the grapevine, pull filings and stock prices. finny organises your analysis in projects, creates dashboards to visualize the data, and stores the backups on your drive.
I run finance for fast growing fintech scale-up. Every week it's comps, benchmarks, model reviews, deal memos — the usual stack of spreadsheets and slides. I'm also a software engineer, and at some point it became obvious. AI models have gotten good enough that the spreadsheet is now the bottleneck.
finny is what happened when I stopped fighting that. It's a financial analysis notebook that runs in your terminal, where both you and the model work in code. Slides become HTML. Spreadsheets become Python and JSON. Analysis that took a morning takes minutes.
The idea is simple: people who learn to work in the machine's native language will be dramatically more effective than those who don't. finny is how you start.
/benchmarking— Compare companies on key metrics with sourced data/trading-comps— Public company multiples table with stats/acquisition-comps— M&A precedent transactions/filings— Pull SEC, ASX, HKSE, LSE filings/market-map— Map companies across an industry landscape/review-model— Analyze Excel models, generate interactive dashboards/grapevine— Reddit, HN, Polymarket, and news signals for a topic
Skills pull live data from these tools automatically. All output JSON — you can also call them directly for scripting.
| Tool | Auth | What it fetches |
|---|---|---|
| Yahoo Finance | None | Quotes, price history, financial statements |
| Financial Modeling Prep | FMP_API_KEY env var |
Profiles, ratios, metrics, peers, estimates |
| Filings | None | SEC EDGAR, ASX, HKSE, LSE regulatory filings |
| Social | None | Reddit, HN, Polymarket discussions |
See tools/ for CLI usage and raw data access.
Track valuations, ARR, revenue growth, and trading multiples across public and private AI companies. finny sources data from yfinance, company filings, and news, then generates an interactive dashboard.
Feed in an Excel model and finny maps the structure, extracts key metrics, and builds a self-contained HTML dashboard with annual/quarterly toggle, revenue drivers, stock price, and trading comps tabs.
Pull growth, margins, and forward earnings for a company and its peers. finny selects comparable companies, fetches live market data via yfinance, and formats a comps table with statistics. Get stock prices and basic financial information about listed companies from Yahoo finance.
Scan Reddit, HN, Polymarket, and news for what practitioners are saying about a topic. finny synthesizes sentiment, key themes, and source links into a research report — injected as a dashboard tab when a project is active.
Pull filings from SEC EDGAR, ASX, HKSE, and LSE. For earnings analysis, finny extracts KPIs and builds side-by-side comparisons with revenue mix breakdowns.
Requirements: Python 3.10+, Claude Code
git clone https://github.com/jyek/finny.git
cd finny
bin/finny setupsetup creates a virtual environment, installs Python dependencies, and symlinks all skills into .claude/skills/ in your finny folder.
finny start # serves at localhost:8787The dashboard shows all your project dashboards with live reload. Create your first project so you can run your first analysis.
You can create new projects or switch projects from the dropdown in the top-right corner.
Open Claude Code in your terminal or a code editor like VSCode. Run any skill or ask it to do an analysis in natural language. For instance:
Get trading comps
/trading-comps Stripe, Adyen, Block
or
Perform benchmarking
/benchmarking --metrics "ARR, NRR, CAC payback" --companies "Rippling, Deel, Workday"
or
Get insights from social media and polymarkets
/grapevine "vertical AI infrastructure"
Data is saved as JSON and Markdown files in your workspace. Visualisations are created on that data.
Go to your dashboard to see the results.
finny stores all project data (comps tables, dashboards, reports) in a workspace folder. By default this is workspace/ inside the repo.
To store projects elsewhere (e.g. Google Drive), set FINNY_HOME in your shell:
export FINNY_HOME="$HOME/Google Drive/My Drive/finny"Add it to your .zshrc or .bashrc to persist across sessions. If FINNY_HOME is not set, finny falls back to workspace/ in the repo root.
Each project gets its own subfolder, and config.json in the workspace root tracks which project is active. Skills automatically save output to the active project.
MIT