1. Does The NM-CYD-C5 Run Ollama Directly?

No. The ESP32-C5 does not have enough memory or computing capacity to run this Qwen model. Ollama, FMP requests, scoring, Telegram polling, and the HTTP API run on the Mac. The CYD is the touchscreen client.

2. Can I Use The CYD Without Keeping It Connected To The Mac?

Yes, after the sketch is uploaded. Power the CYD from a USB power bank and connect it to Wi-Fi. The Mac does not need a USB connection, but it must remain on, awake, online, and running Ollama plus bot.py.

3. Can I Use The Telegram Bot Away From Home?

Yes. Your phone can use mobile data or another Wi-Fi network. Telegram routes the message to the Python bot on your home Mac. The Mac must remain connected to Telegram and the internet.

4. Is The Fundamental Score Generated By AI?

No. Python applies transparent thresholds in scoring.py and produces the score before Ollama is called. Ollama receives the verified result and writes a short interpretation. It cannot alter the score used by Telegram or the CYD.

5. Why Should Analyst Price Targets Not Affect The Score?

Price targets describe external expectations rather than the company’s underlying profitability, financial health, valuation, or cash generation. The application shows their averages and coverage counts as sentiment context but assigns them zero fundamental points.

6. Can I Add More Data Or Change The Criteria?

Yes. Add endpoints in fmp_client.py, define transparent rules in scoring.py, update the report pages in analysis_service.py, and extend the Ollama prompt only after the deterministic calculations are complete. Strong next steps include multi-year trends, sector peer comparisons, earnings growth, dilution, and sector-specific models for financial companies and REITs.

Beyond A Stock Ticker: Build An ESP32 AI Stock Analyzer

Sanzhi Kobzhan

Sep 23

14 min read

Updated: Sep 25

DIY ESP32 AI stock analyzer running on an NM-CYD-C5 touchscreen

Build a DIY ESP32 stock analyzer that turns one ticker into a structured fundamental report on an NM-CYD-C5 touchscreen and in Telegram.

Unlike a basic ESP32 stock ticker that only displays price, this system retrieves current financial data, calculates a transparent score, identifies strengths and risks, and uses local Ollama to explain the verified result.

This beginner guide covers the complete build, from installation to first analysis (Code, calculation logic and a step-by-step guide are included).

Key Takeaways

One ESP32 stock-analysis engine serves two interfaces: a Telegram bot on your phone and an NM-CYD-C5 touchscreen powered by USB or a power bank.
Python calculates the 0–100 fundamental score from financial data; Ollama only explains the verified result and cannot change the rating.
The CYD stores no FMP or Telegram secrets. Those remain in a private .env file on the Mac.
The finished project evaluates quality, financial health, valuation, cash generation, and analyst-target context while clearly reporting its limitations.

Why I Built This Project

I worked as an investment analyst for many years across investment banks, second-tier banks, and brokerage companies. During that time, I analyzed listed businesses, financial statements, valuation multiples, balance-sheet risk, and market expectations. I also built several tools for equity investors, including the Stocks2Buy platform and app.

More recently, I started learning electronics. The NM-CYD-C5 immediately suggested a useful experiment: could I combine my finance background with a compact ESP32 touchscreen and a local language model? The result moves beyond a novelty stock-market display. It is a working fundamental-analysis terminal with the same output available through Telegram.

The project is designed for traders, investors, investment analysts, and electronics beginners. It demonstrates APIs, Python services, local AI, secure configuration, Wi-Fi communication, ESP32 programming, touch interfaces, and practical debugging in one build.

Final device demo

Device video demo

https://video.wixstatic.com/video/b34654_a2540b54cf7844919f5d1b9ccd1bed35/720p/mp4/file.mp4

NM-CYD-C5 stock analyzer

What This AI Stock Analyzer Does

The user enters a ticker such as AAPL, MSFT or KO. The Mac requests three datasets from the API endpoint, applies a fixed scoring model, and sends the verified data to Ollama for a short interpretation.

The report contains a 0–100 score, an overall conclusion, data coverage, quality, leverage, liquidity, valuation and cash-flow metrics, rule-based strengths and risks, unscored analyst-target context, an Ollama interpretation, and explicit limitations.

Telegram and the CYD call the same AnalysisService, so the calculation is consistent across both interfaces. A ten-minute cache also prevents the same ticker from unnecessarily consuming another full set of API requests. The result combines a DIY stock ticker, a Telegram stock bot, and local AI stock analysis in one practical build.

How The ESP32 Stock Analyzer Works

Component	Responsibility
API endpoint provider.	Supplies current TTM metrics, ratios, and price-target summaries
Python service on the Mac	Validates tickers, requests financial data, calculates the score, caches results, and serves both interfaces
Ollama	Runs the local analyst model and explains the verified result
Telegram bot	Accepts tickers remotely and returns the complete report
NM-CYD-C5	Provides Wi-Fi setup, an on-screen ticker keyboard, and paginated results

The CYD is a client, not the server. After flashing, it can run from a power bank without a USB connection to the Mac. The Mac must remain on, awake, online, and running Ollama plus bot.py.

Telegram works differently. Your phone can be outside the house and connected through mobile data because Telegram carries the message to the bot process running on the Mac. The Mac still has to remain online.