Indian Equity Sector Performance Analysis

A SQL + Python (Streamlit) analysis of how different sectors of the Indian stock market (NSE) have performed over time — built as a portfolio project for Finance Analyst / Business Analyst roles.

Rather than just plotting stock prices, this project asks specific business questions of the data: which sectors led, which lagged, how did each sector behave during market corrections, and how correlated are sectors with the broader market. The analytical work happens in SQL; the results are visualized in an interactive Streamlit dashboard.

New to this project / setting it up for the first time? Use SETUP_CHECKLIST.md for a linear, copy-paste checklist (Mac-focused) that also covers the most common setup errors.

---

Why this project

Most "stock dashboard" projects show a chart and stop there. This one is built to demonstrate the actual skill set a Finance/Business Analyst role needs:

• Translating a business question into a query
• Working confidently in SQL (window functions, aggregations, correlation)
• Building a clear, decision-oriented dashboard rather than a generic chart
• Writing up findings as an analyst would, not just generating numbers

---

Business questions answered

1. Which sectors delivered the best risk-adjusted returns over the period?
2. Which individual stocks led and lagged within each sector?
3. How far did each sector fall (drawdown) during a market correction, and which sectors were most defensive?
4. Which sectors move closely with the broader market (Nifty 50), and which are more independent — useful for diversification thinking?
5. Does sector leadership rotate over time, and is there a visible pattern?

---

Project structure

.
├── data/
│   ├── raw/              # Raw CSVs downloaded from NSE (gitignored, regenerate locally)
│   └── processed/        # Cleaned CSVs ready for SQL load (gitignored, regenerate locally)
├── scripts/
│   ├── fetch_data.py     # Downloads historical stock + sector index data from NSE
│   └── clean_data.py     # Cleans raw data, computes daily returns, writes processed CSVs
├── sql/
│   ├── schema.sql            # Table definitions
│   ├── load_data.sql         # Loads processed CSVs into the database
│   └── analysis_queries.sql  # The analytical core -- one query per business question
├── dashboard/
│   ├── app.py             # Streamlit dashboard (3 pages: Overview, Sector Deep Dive, Rotation & Correlation)
│   └── db.py              # Database connection helper (reads credentials from .env)
├── docs/
│   └── INSIGHTS_TEMPLATE.md  # Template for writing up your findings once you have results
├── requirements.txt
├── .env.example           # Template for database connection variables
├── SETUP_CHECKLIST.md      # Linear copy-paste setup guide + common error fixes
└── README.md

---

Tech stack

Layer	Tool
Data source	NSE historical data via the jugaad-data Python library
Data cleaning	Python (pandas)
Storage / analysis	SQL (PostgreSQL syntax; minor edits work for SQL Server/MySQL/SQLite)
Visualization	Streamlit + Plotly (interactive Python dashboard)

No paid API keys are required — jugaad-data reads NSE's own public historical data archives.

---

How it works, end to end

1. Fetch raw data

pip install -r requirements.txt
python scripts/fetch_data.py

This downloads daily OHLC data for a basket of ~25 NSE stocks spanning 8 sectors (Banking, IT, Pharma, Auto, FMCG, Energy, Infrastructure, Metals, Telecom), plus the corresponding sector indices (Nifty Bank, Nifty IT, etc.) and the Nifty 50 benchmark. Edit STOCK_UNIVERSE and SECTOR_INDICES in scripts/fetch_data.py to change the universe or date range. Files land in data/raw/.

2. Clean and prepare the data

python scripts/clean_data.py

This standardizes column names/types, sorts by date, and computes daily percentage returns for every stock and index. Output lands in data/processed/ as three files: daily_prices.csv, sector_index_prices.csv, and stock_master.csv (the sector mapping).

3. Load into a SQL database

# create the tables
psql -d your_database -f sql/schema.sql

# load the cleaned CSVs
psql -d your_database -f sql/load_data.sql

(If you're using a GUI tool like pgAdmin or DBeaver instead of psql, run schema.sql there and use the tool's CSV import wizard pointed at the files in data/processed/ — same result, different interface.)

4. Run the analysis

Open sql/analysis_queries.sql and run each query against your database. Each one is commented with the business question it answers. These are also the queries the Streamlit dashboard's caching layer mirrors.

5. Run the Streamlit dashboard

# create a .env file with your DB credentials (copy from .env.example)
cp .env.example .env

streamlit run dashboard/app.py

This opens an interactive dashboard in your browser with three pages:

• Market Overview — Nifty 50 trend, sector returns heatmap, risk-vs-return scatter
• Sector Deep Dive — pick a sector, see its stocks' price trends and a ranked performance table
• Sector Rotation & Correlation — each sector's correlation with Nifty 50, and which sector led each month

All charts respect a date-range filter in the sidebar.

6. Write up your findings

Use docs/INSIGHTS_TEMPLATE.md to document what you found. This write-up is arguably the most important deliverable for a resume project — it's the narrative you'll actually talk through in interviews.

---

Data scope and assumptions

• Stock universe: ~25 stocks across 8 sectors, chosen to be representative rather than exhaustive (full Nifty 50 coverage adds fetch time without changing the analytical approach).
• Returns: Price returns only — dividends are not factored in, so total return figures are a slight underestimate of true investor returns. This is noted explicitly in the insights template.
• Frequency: Daily closing-price based. Intraday volatility isn't captured.
• Corporate actions: jugaad-data pulls NSE's series data, which generally reflects adjusted prices for standard corporate actions, but it's worth spot-checking any stock that had a major split/bonus during your date range if a return number looks unusually large.

---

Possible extensions

• Pull fundamentals (P/E, market cap) per stock and add valuation-based analysis alongside the return/volatility analysis
• Extend to the full Nifty 50 / Nifty 500
• Add dividend-adjusted total return calculations
• Automate periodic refresh with a scheduled script instead of one-time historical pull
• Add a macro overlay (repo rate changes, FII/DII flow data) to explain why certain sectors moved, not just that they moved

---