Testing Betting Against Beta

B.Sc. thesis.

University of Konsanz Wimter 2025/2026

Replication of the BAB factor (Frazzini & Pedersen, 2014) for U.S. equities, January 2000 to December 2020.

Research Question

Why is the empirical Security Market Line flatter than the CAPM implies, which market frictions generate this flattening, and can these distortions be systematically exploited through a Betting Against Beta framework?

The analysis tests three propositions derived from the Margin-CAPM:

1. CAPM alphas decline from low-beta to high-beta decile portfolios (flat SML).
2. A beta-neutral BAB factor earns significant risk-adjusted returns across factor models.
3. BAB returns deteriorate when funding liquidity conditions tighten.

How to Run

Make sure you're inside the project folder, then:

pip install numpy pandas statsmodels scipy matplotlib openpyxl requests tqdm
python bettingagainstbeta.py

That's it. Factor data (Fama-French, momentum, q-factors, TED spread) downloads automatically on first run. Results and figures go into data/ and data/figures/.

If you want to rebuild the panel from scratch (not needed for the analysis):

python databank-download.py --download   # needs EODHD API key, takes hours
python databank-prepare.py               # ~15 min
python databank-filter.py

Data

Source	What	Required
EODHD	Daily U.S. equity prices	Yes (via download script)
Kenneth French Data Library	FF3, FF5, momentum, risk-free rate	Yes (auto-downloaded)
global-q.org	q-factor model	Optional (auto-downloaded)
FRED	TED spread	Optional (auto-downloaded)
AQR	Published BAB returns (benchmark)	Optional (manual download)

The AQR file goes into data/ as Betting_Against_Beta_Equity_Factors_Monthly.xlsx. If missing, the benchmark comparison is skipped.

Project Structure

├── bettingagainstbeta.py       Main analysis, all results and figures
├── databank-download.py        EODHD download (optional, resumable)
├── databank-prepare.py         Daily prices → monthly panel + betas
├── databank-filter.py          Sample filters
├── databank-audit.py           Data quality checks
└── data/                       All inputs, outputs, and figures

AI Usage

I used AI tools for coding assistance (syntax, debugging). The whole elaboration, methodological approach, implementation, and interpretation of results are entirely my own.