Python Environment, Core Syntax & My First Finance Script

Why Miniconda and Not Just pip

Before writing a single line of Python, I spent time getting the environment right. On an M4 Mac, this matters more than on a standard x86 machine — PyTorch in particular needs the right build for Apple Silicon to use MPS (Metal Performance Shaders) for GPU acceleration. A wrongly installed torch on M4 runs on CPU only and you'll never know it.

I chose Miniconda over plain pip for one reason: environment isolation. Every project gets its own environment with its own Python version and its own package versions. When you're working across Phase 2 (ML), Phase 3 (PyTorch), and Phase 5 (LangChain), these packages can conflict badly. Conda solves that cleanly.

Setting Up Miniconda on M4 Mac

This is the exact sequence I ran. If you're on an M1/M2/M3/M4 Mac, use the arm64 installer specifically — not the x86 one.

# Download the arm64 installer for Apple Silicon
curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-arm64.sh

# Run the installer
sh Miniconda3-latest-MacOSX-arm64.sh

# Initialise conda for zsh (default shell on M4 Mac)
~/miniconda3/bin/conda init zsh

# Reload shell config
source ~/.zshrc

# Verify conda is working
conda --version
conda 24.11.1

One thing that caught me — after running conda init zsh you must restart the terminal or run source ~/.zshrc. If you skip this, the conda command won't be found.

Accepting the Terms of Service

Conda now requires explicitly accepting the Terms of Service for the default channels. Without this, package installs fail with a cryptic error:

conda tos accept --override-channels --channel https://repo.anaconda.com/pkgs/main
conda tos accept --override-channels --channel https://repo.anaconda.com/pkgs/r

Creating the AI Dev Environment

# Create a dedicated environment — Python 3.11 specifically
conda create -n ai_dev python=3.11 -y

# Activate it — you'll see (ai_dev) in your prompt
conda activate ai_dev

# Install the Day 1 libraries
pip install numpy pandas pydantic torch yfinance matplotlib

# Verify torch is using MPS (Apple Silicon GPU)
python3 -c "import torch; print(torch.backends.mps.is_available())"
True

Core Python Syntax — What I Focused On

The goal today was not to master Python. It was to become comfortable reading and writing the patterns that appear constantly in data and AI work. I did not touch decorators, metaclasses, async/await, or anything complex. Those come when they're needed.

Here's what I actually wrote in basics.py — with financial data examples throughout, not generic toy examples:

# ── Variables & Data Types ───────────────────────────────
num = 3
num2 = 4
Name = "Prabhu"
age = 38

# Type conversion — useful when dealing with API responses
num_str = str(num)
print(num_str)                    # "3"
print(str(num + num2) + Name)    # "7Prabhu"
print(f"my age is {age}")        # f-strings — use these everywhere

# ── Lists — building a stock watchlist ──────────────────
stocks = []
stocks.append("AAPL")
stocks.append("TSLA")
stocks.append("MSFT")
stocks.insert(1, "HDFC")     # Indian BFSI stock at position 1
stocks.insert(2, "JIO")
stocks.append("Rakbank")
stocks.pop()                   # removes last item
stocks.pop(1)                 # removes item at index 1

print(f"stocks: {stocks}")
print(f"length: {len(stocks)}")

# Loop through watchlist
for stock in stocks:
    print(stock)

# ── Dictionaries — stock price map ───────────────────────
stock = {}
stock["AAPL"] = 200
stock["HDFC"] = 300
stock.update({"JIO": 300})

del stock["AAPL"]

# Insert at specific position (dict doesn't support insert natively)
items = list(stock.items())
items.insert(1, ("TCS", 400))
stock = dict(items)

print(f"portfolio: {stock}")
stock.clear()

# ── Functions ────────────────────────────────────────────
def greet(name):
    return f"Hello {name}"

print(greet(Name))

The Day 1 Project — AAPL Stock Analyser

This is the real output of Day 1. A script that downloads 5 years of Apple stock data, runs basic analysis, exports a CSV, and saves a price chart as a PDF. Small? Yes. But every step is real — real library, real data, real file output.

import yfinance as yf
import pandas as pd
import matplotlib.pyplot as plt

# ── Explore the Ticker object ────────────────────────────
# yf.Ticker gives you everything about one stock
stock = yf.Ticker("AAPL")

# These all work — try uncommenting one at a time
# print(stock.financials)    # Income statement
# print(stock.info)          # Company overview
# print(stock.news)          # Recent news
# print(stock.balance_sheet) # Balance sheet
# print(stock.cashflow)      # Cash flow statement
# print(stock.dividends)     # Dividend history

# ── Download historical price data ──────────────────────
data = yf.download("AAPL", start="2020-01-01")

# Uncomment to explore the data structure
# print(data.head())                    # First 5 rows
# print("Max Close:", data["Close"].max())
# print("Min Close:", data["Close"].min())
# print("Avg Close:", data["Close"].mean())

# Daily returns — how much did price change each day?
# data["Daily Return"] = data["Close"].pct_change()
# print(data["Daily Return"].head())

# ── Export to CSV ────────────────────────────────────────
data.to_csv("aapl_stock_data.csv")
print("CSV file saved successfully")

# ── Plot the close price ─────────────────────────────────
data["Close"].plot()
plt.savefig("chart.pdf", dpi=300)
# plt.show()  # Uncomment to display interactively

What Each Part Does

News Data — What I Explored

I also explored the yFinance news API — not in the final script but in a commented block. The structure is worth understanding because it's your first encounter with nested JSON parsing, which appears constantly in AI API responses:

# This pattern appears in EVERY AI API response
# Learn it now — it will save you hours later

for story in news_data:
    content = story.get('content', {})     # .get() never raises KeyError
    title   = content.get('title', 'No Title')
    summary = content.get('summary', 'No Summary')

    print(f"TITLE: {title}")
    print(f"SUMMARY: {summary}")
    print("-" * 30)

📊 Actual Output chart.pdf → chart.png — generated by day1_stock_project.py

This is the actual chart produced by the script on Day 1 — AAPL closing price from 2020 to 2026, generated with matplotlib and saved as a PDF at 300 DPI.

💡 SAVING IN DIFFERENT FORMATS

The script saves as PDF — great for print and presentations. You can save in any format just by changing the filename extension:

plt.savefig("chart.png") plt.savefig("chart.jpg") plt.savefig("chart.svg") plt.savefig("chart.pdf", dpi=300)

PNG — best for web and dashboards · JPG — smaller file, slight quality loss · SVG — vector, scales infinitely · PDF — vector, best for reports. The dpi parameter only affects raster formats (PNG, JPG) — higher = sharper.

What I Did NOT Do Today

This is as important as what I did do. It's very easy to go down rabbit holes on Day 1.

• No ML models — that's Phase 2 (Day 11)
• No Pandas deep dive — that's Days 7–8
• No NumPy — that's Day 3
• No advanced OOP, decorators, generators
• No 4-hour YouTube tutorials
• No jumping ahead to LangChain or PyTorch

The roadmap is sequenced deliberately. Everything introduced today — variables, lists, dicts, loops, functions, yfinance — is used in every subsequent phase. Rushing ahead before these are solid creates confusion later.

The Writing Prompt — Why I'm Doing This

The roadmap requires a 1-paragraph writing reflection each day. Today's prompt: "Why does clean environment setup prevent 'works on my machine' issues?"

What I Built by End of Day 1

What's Coming on Day 2

Day 2 covers Python OOP + File I/O + Poetry. I'll build a FinancialRecord class that loads the CSV we exported today, stores records as objects, and exposes a summary() method. I'll also set up Poetry — the modern dependency management tool that replaces bare pip for production projects.