AI Agent for Financial Analysis

This is a small Python tool that helps you quickly create nice financial reports.

Summary – script overview

File	Main job	When it runs
agent.py	Runs everything, coordinates steps	First (you start here)
download_clean.py	Downloads + basic cleaning	Step 1
data_cleaner.py	Cleaning rules	Called by download_clean
eda_comparative.py	Summary stats + correlation matrix	Step 2
feature_engineering.py	Adds technical indicators	Step 3
feature_functions.py	The actual indicator calculations	Called by feature_engineering
viz_comparative.py	Creates comparison charts (multi-asset)	Step 5
report_generator.py	Builds the final HTML report	Last step
eda_functions.py	Single-asset EDA helpers (not main flow)	Optional / legacy
viz_functions.py	Single-asset chart helpers (not main flow)	Optional / legacy

All these files work together so that you only need to write a sentence and get a complete financial comparison report without touching the code.

How it works (simple explanation)

Behind the scenes, the program follows these main steps:

1. You write your question in normal words
   (example: "compare gold and silver" or "top 5 tech stocks performance")

2. A small local AI model (Qwen 2.5 3B running with Ollama) reads your sentence
   → understands what you want
   → chooses which companies or commodities to analyze

3. The program automatically downloads fresh price data
   → uses Yahoo Finance (yfinance library)

4. It cleans the data
   → removes errors, fixes formats, makes sure dates and numbers are correct

5. It does some quick analysis (EDA)
   → calculates basic statistics
   → creates a correlation matrix between assets

6. It calculates useful technical indicators
   → RSI (overbought / oversold)
   → Bollinger Bands
   → rolling volatility
   → cumulative returns
   → moving averages, MACD, etc.

7. It creates clear comparison charts
   → price trends side by side
   → cumulative performance
   → RSI comparison
   → volatility comparison
   → Bollinger Bands width

8. Finally, it builds a nice HTML report
   → summary table for each asset (min / median / max)
   → small preview chart for each asset
   → big comparison charts
   → everything opens in your browser

All of this happens in a few minutes, and every new request creates its own separate folder so nothing gets mixed up.

You don't need to write any code or open spreadsheets — just describe what you want to see.

Everything happens automatically — no need to open Excel or TradingView manually.

What you can do with it

• Compare prices and performance of several stocks or commodities at once
• See which asset is more volatile right now
• Look at RSI, cumulative returns, Bollinger width side by side
• Get a clean report you can save or share

Quick examples of what you can ask

• "I want a report on the top 5 commodities"
• "Compare NVDA, AAPL, MSFT and TSLA"

How to try it

1. Download or clone this repository
2. Install Python (if you don't have it already — version 3.9 or newer is fine)
3. Double-click (or run) the file called src/agent.py
4. When a box appears, just type your question and press Enter

The program will create a new folder inside results/analyses/ with all the charts and the final HTML report.

You can open the report by double-clicking the file report.html — it should open in your browser.

See some ready-made examples

Inside the folder results/examples/ you can already find a few nice-looking reports:

• Major commodities (gold, silver, oil, gas, copper)
• Top US tech stocks
• Energy sector vs technology sector

Just open any report.html file there to see how the final result looks.

Project folders explained

• src/ → all the Python code files
• results/ → where new reports are saved
• results/examples/ → some beautiful ready-to-view reports

More details:

• README-src.md → what each Python file does
• README-results.md → how to understand and use the generated reports

Made with

• Python
• yfinance (to get market data)
• pandas, matplotlib, seaborn (for tables and charts)
• ollama qwen 2.5:3b as a LLM to help to understand the prompt

Enjoy exploring markets visually and quickly!