BankLab: JPM vs Morgan Stanley Analytics Platform

A reproducible research platform for comparative analysis of JPMorgan Chase (JPM) and Morgan Stanley (MS) using public data sources.

Overview

BankLab provides a clean, professional pipeline for:

• Fundamentals: SEC EDGAR XBRL company facts (10-K/10-Q filings)
• Market Data: Daily prices and returns from public sources
• Factor Models: Fama-French 5-factor daily returns
• Macro Context: FRED economic indicators

Quick Start

# Clone and setup
git clone https://github.com/Leotaby/banklab.git
cd banklab
pip install -e ".[dev]"# Set FRED API key (get free key at https://fred.stlouisfed.org/docs/api/api_key.html)export FRED_API_KEY="your_key_here"# Download and process all data
make data
# Run tests
make test# Build full report
make report

Data Sources

Source	Description	Update Frequency
SEC EDGAR	Company facts, filings metadata	Quarterly
Stooq	Daily stock prices	Daily
Fama-French	5-factor model returns	Daily
FRED	Fed Funds Rate, Treasury yields, GDP	Various

Project Structure

banklab/
├── src/banklab/ # Main package
│ ├── ingest/ # Data downloaders
│ │ ├── sec.py # SEC EDGAR API
│ │ ├── market.py # Stock prices
│ │ ├── factors.py # Fama-French factors
│ │ └── macro.py # FRED macro series
│ ├── process/ # Data transformations
│ │ └── pipeline.py # Processing pipeline
│ ├── utils/ # Shared utilities
│ │ ├── cache.py # Caching + manifest
│ │ └── http.py # Polite HTTP client
│ └── run.py # CLI entry point
├── tests/ # pytest test suite
├── data/
│ ├── raw/ # Cached downloads (gitignored)
│ └── processed/ # Output parquet files
├── notebooks/ # Jupyter analysis notebooks
├── docs/ # Documentation
└── data_manifest.yml # Data provenance log

Output Schemas

prices_daily.parquet

Column	Type	Description
date	date	Trading date
ticker	string	Stock ticker (JPM, MS)
close	float64	Adjusted close price
ret	float64	Daily return

factors_daily.parquet

Column	Type	Description
date	date	Trading date
mktrf	float64	Market excess return
smb	float64	Size factor
hml	float64	Value factor
rmw	float64	Profitability factor
cma	float64	Investment factor
rf	float64	Risk-free rate

macro_monthly.parquet

Column	Type	Description
date	date	Observation date
series_id	string	FRED series identifier
value	float64	Observation value

fundamentals_raw_facts.parquet

Column	Type	Description
date	date	Filing/period end date
cik	string	SEC CIK identifier
ticker	string	Stock ticker
tag	string	XBRL taxonomy tag
value	float64	Reported value
unit	string	Unit of measure
fp	string	Fiscal period (FY, Q1-Q4)
fy	int64	Fiscal year
form	string	Filing form type

Makefile Commands

make data # Download and process all data
make build # Install package in dev mode
make test# Run pytest suite
make lint # Run ruff linter
make report # Generate analysis report
make clean # Remove cached data
make all # Full pipeline: build → data → test → report

Configuration

Environment Variables

Variable	Required	Description
FRED_API_KEY	Yes	FRED API key for macro data
BANKLAB_DATA_DIR	No	Override default data directory

Rate Limiting

The SEC requires polite access with proper User-Agent headers. BankLab implements:

• 10 requests/second max to SEC EDGAR
• Exponential backoff on 429 responses
• Persistent caching to minimize repeated requests

Development

# Install with dev dependencies
pip install -e ".[dev]"# Run linter
ruff check src/ tests/
# Run tests with coverage
pytest --cov=banklab --cov-report=term-missing
# Format code
ruff format src/ tests/

License

MIT License - see LICENSE for details.

Citation

If you use this platform in research, please cite:

@software{banklab2025,
author = {Hatef Tabbakhian},
title = {BankLab: JPM vs Morgan Stanley Analytics Platform},
year = {2025},
url = {https://github.com/Leotaby/banklab}
}