Chat with Fundamentals v1.0 — Local AI Equity Research Platform

Open-source and local-first — released Tuesday, April 29.

Due to the growth of the project, Chat with Fundamentals has now been separated into a standalone application.

The core idea is simple: generate an LLM-based financial analysis grounded in quantitative elements — namely OHLCV data, financial news, and fundamental metrics. All of this data is sourced from EODHD APIs.

The user enters a chat query, which is parsed to determine which data to fetch.
Currently, the AI flow is designed as a one-shot process — it does not maintain a conversation history after the initial analysis.
This approach ensures focus on single queries about one or several tickers.

Chat with Fundamentals v1.0 is open-sourced under the Apache 2.0 License. Users are free to modify, extend, and deploy it locally.

---

Typical Use Cases

• "Is TSLA a good buy?"
  → Fetches the latest EOD data, recent news, and key financial metrics for TSLA.
  → The LLM generates a concise executive summary.
  → Raw OHLCV data and JSON inputs can also be saved for further research.
  → A “Know More” button gives access to:

  • A TradingView-style financial chart

  • A Monte Carlo simulation of forward returns

  • Value at Risk (VaR) calculation

  • Distribution of daily returns

  • Correlation scatter plot against a benchmark

  • 3-year cumulative return comparison with the benchmark

Test ‘Is TSLA a good buy ?’ - first page

Second page with OHLCV Dataframe, chart and access to JSON raw input

• "Compare TSLA and AMZN"
  → Fetches data and metrics for both tickers.
  → The LLM builds a comparative executive summary based on fundamentals, news, and price action.

  

  Test ‘Compare TSLA and AMZN’

  Disclosure: I am affiliated with EODHD. Subscribing via my links helps support the project at no additional cost to you.

---

Built-in Analytics

Beyond natural language summaries, Chat with Fundamentals provides a foundation for deeper quantitative analysis.
The Quant Analysis module already includes:

• Monte Carlo simulations of future equity paths

• Value at Risk (VaR) and Expected Shortfall (ES) estimates

• Return distribution histograms

• Beta and R² scatter plots against benchmarks

• Cumulative return comparisons over 3 years

The backend is structured to allow progressive extension, with future upgrades such as:

• Rolling Sharpe ratios and volatility bands

• Calendar heatmaps of returns

• Autocorrelation (ACF/PACF) plots

• Volatility cones

• Drawdown and underwater curve visualizations

Security analytics - no-LLM workflow

---

Local-First Model and Future Evolutions

This application is designed to be local-first. It is intended to be deployed on your own machine, with a setup script and desktop launcher provided.

I strongly believe that the future of AI lies in owning your data pipelines, building services locally, and tailoring workflows to your needs.
Technical barriers have fallen — individuals and small teams can now develop platforms that fit their risk profiles, domain knowledge, and specific goals.

In this spirit, the released code is meant to be a boilerplate.
Natural evolutions of the project include:

• Using fine-tuned local models (subject to licensing terms)

• Expanding data integrations

• Improving executive research workflows

---

Disclaimer

This tool is designed to support, not to replace, personal judgment.
Users are expected to exercise due diligence and double-check LLM outputs.
Nothing in this application or associated research should be considered an invitation to buy or sell any security.

S.M.L.