Paper: arXiv 2409.00416

Authors: Miguel C. Herculano

Abstract

Frazzini and Pedersen (2014) Betting Against Beta (BAB) factor is based on the idea that high beta assets trade at a premium and low beta assets trade at a discount due to investor funding constraints. However, as argued by Campbell and Vuolteenaho (2004), beta comes in “good” and “bad” varieties. While gaining exposure to low-beta, BAB factors fail to recognize that such a portfolio may tilt towards bad-beta. We propose a Betting Against Bad Beta factor, built by double-sorting on beta and bad-beta and find that it improves the overall performance of BAB strategies though its success relies on proper transaction cost mitigation.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper involves advanced theoretical finance math (ICAPM, VAR decompositions, precise beta definitions) but is heavily grounded in empirical backtesting with CRSP data, transaction cost analysis, and replication of existing factors, showing a blend of theoretical and practical rigor.

Research Flowchart

  flowchart TD
  subgraph A ["Research Goal"]
      G1["Does ignoring 'bad beta'<br>limit Betting Against Beta<br>performance?"]
  end

  subgraph B ["Key Methodology"]
      M1["Database: CRSP & Compustat<br>Data: US Stocks, 1963-2013"]
      M2["Double Sort Strategy<br>1. Sort by Market Beta<br>2. Sort by Bad Beta (Covariance with<br>bad macro states)"]
  end

  subgraph C ["Computational Process"]
      C1["Reconstruct Standard BAB<br>(Frazzini & Pedersen 2014)"]
      C2["Construct BABB: Betting Against<br>Bad Beta Factor<br>(Long Low Bad-Beta / Short High Bad-Beta)"]
  end

  subgraph D ["Key Findings & Outcomes"]
      F1["Performance:<br>BABB generates higher risk-adjusted<br>returns than standard BAB"]
      F2["Limitation:<br>Success depends on strict<br>transaction cost mitigation"]
      F3["Conclusion:<br>Controlling for Bad Beta improves<br>BAB strategies"]
  end

  A --> B
  B --> C
  C --> D

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes a Betting Against Bad Beta (BABB) factor that double‑sorts on beta and bad‑beta to improve the performance of the original BAB strategy.

Source sentence
We propose a Betting Against Bad Beta factor, built by double-sorting on beta and bad-beta and find that it improves the overall performance of BAB strategies though its success relies on proper transaction cost mitigation.

Assumes 6

  • The economy follows a first‑order vector‑autoregressive (VAR) model. model
    Source sentence
    Assume that the economy is well described by a first-order vector-autoregressive (VAR) model of the form xt+1 = µx + Γxt + σut+1.
  • The VAR is homoskedastic (σ is time‑invariant). statistical
    Source sentence
    In general, an homoskedastic VAR is assumed and therefore σ in (6) is time-invariant.
  • The log‑linearization parameter ρ in the Campbell‑Shiller identity is set to 0.95. model
    Source sentence
    ρ is a log-linearization parameter in the Campbell‑Shiller identity, set to 0.95 by most authors.
  • Four state variables (excess market return, yield spread, CAPE, small‑stock value spread) capture the relevant information for cash‑flow and discount‑rate news. data
    Source sentence
    We follow Campbell and Vuolteenaho (2004) and include four state variables in the model: excess stock market returns rt
Show 2 more
  • Leverage the low‑beta portfolio and deleverage the high‑beta portfolio to achieve market neutrality. execution
    Source sentence
    Similar to the BAB of Frazzini and Pedersen (2014), the BABB is constructed by leveraging the low beta portfolio and deleveraging the high beta portfolio, with a view of achieving market neutrality.
  • Transaction costs are approximated by the simple average of four low‑frequency firm‑level bid‑ask spread measures. execution
    Source sentence
    We take after Chen and Velikov (2023); Novy‑Marx and Velikov (2024) and consider the simple average of four low‑frequency (LF) measures of transaction costs, based on half the effective firm‑level bid‑ask spread.

Data

Backtest sample: 1963-01 – 2021-01 · US stocks (CRSP) · equities · monthly

  • CRSP US stock universe · CRSP market-prices public
    1963‑01 to 2021‑12 · US stocks, equal‑weighted, sorted into thirds
    Source sentence
    The BAB is long‑short the bottom‑top thirds of equal‑weighted US stocks in CRPS, sorted on the basis of the Frazzini‑Pederson betas from 1963:01-2021:12.
  • Low‑frequency transaction‑cost measures · Various (Hasbrouck, Corwin & Schultz, Abdi & Ranaldo, Kyle & alternative public
    Source sentence
    consider the simple average of four low‑frequency (LF) measures of transaction costs, based on half the effective firm‑level bid‑ask spread. Specifically, we consider Hasbrouck (2009) Gibbs estimate, Corwin and Schultz (2012) high‑low spread, Abdi and Ranaldo (2017) close‑high‑low spread, and Kyle and Obizhaeva (2016)
  • State variables for VAR · CRSP (market data) and macro sources macro public
    Source sentence
    the yield spread between 10 year and 3 months Treasury Bonds, the Cyclically Adjusted Price Earning Ratio (CAPE) and the small‑stock value spread.

Costs

Modeled Costs are part of the reported headline results.

Costs are incorporated in net returns using the average of four low‑frequency bid‑ask spread measures (Gibbs, high‑low, close‑high‑low, VoV).

Source sentence
consider the simple average of four low‑frequency (LF) measures of transaction costs, based on half the effective firm‑level bid‑ask spread.

Finds 5

Reported performance (Betting Against Bad Beta (BABB) factor): Sharpe 1.09 (benchmark 1.01) · annual return 15.0%

  • The BABB factor has a gross mean annualized return of 15.0% and volatility of 13.8%, compared with 11.4% and 11.3% for the BAB. performance
    Source sentence
    The Betting Against Bad Beta (BABB) factor has a gross mean annualized return of 15.0% and a annualized volatility of 13.8% which compares with 11.4% and 11.3% of BAB, respectively.
  • BABB’s Sharpe ratio is 1.09, about 800 bps higher than BAB’s Sharpe of 1.01 for the sample period. performance
    Source sentence
    The Sharpe ratio of the BABB is 1.09, approximately 800bps higher compared with the BAB which is 1.01 for this specific sample period and construction scheme.
  • BABB’s five‑factor alpha is 75 bps per month, substantially higher than BAB’s 51 bps per month. performance
    Source sentence
    In particular, we find that the BABB has a five‑factor alpha of 75 bps per month, which compares to
  • Exposure to the five Fama‑French factors explains at most 11% of BABB returns versus about 25% for BAB. statistical
    Source sentence
    First, exposure to these factors explain at most 11 per cent of BABB returns, which contrasts with roughly 25 per cent in the case of BAB.
  • BABB trading costs are significantly higher than BAB’s, with average firm‑level costs around 0.003‑0.005 versus 0.0017‑0.0023 for BAB. performance
    Source sentence
    We find that BABB factor trading costs are significantly higher

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 26 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (36)

  1. Structural VARs and noninvertible macroeconomic models (2019) — link
  2. Where do Betas Come From? Asset Price Dynamics and the Sources of Systematic Risk (1993) — link
  3. An Intertemporal Capital Asset Pricing Model (1973)
  4. Does Information Help Recovering Structural Shocks from Past Observations? (2006)
  5. Growth or glamour? Fundamentals and systematic risk in stock returns (2010) — link
  6. What Drives Booms and Busts in Value? (2023)
  7. Return decomposition (2009) — link
  8. Sufficient information in structural VARs (2014) — link
and 28 more
  1. Common shocks in stocks and bonds (2021) — link
  2. Hard times (2013) — link
  3. An intertemporal CAPM with stochastic volatility (2018) — link
  4. A Variance Decomposition for Stock Returns (1991) — link
  5. Stock Prices, Earnings, and Expected Dividends (1988)
  6. The Dividend-Price Ratio and Expectations of Future Dividends and Discount Factors (1988) — link
  7. Return decomposition (2009) — link
  8. Sufficient information in structural VARs (2014) — link
  9. Bad Beta, Good Beta (2004) — link
  10. Large Bayesian vector auto regressions (2010) — link
  11. A Stock Return Decomposition Using Observables (2022) — link
  12. Structural Vector Autoregressive Analysis (2017)
  13. FRED-MD: A Monthly Database for Macroeconomic Research (2016)
  14. Betting against beta (2014) — link
  15. Betting against betting against beta (2022) — link
  16. Zeroing In on the Expected Returns of Anomalies (2023)
  17. Assaying Anomalies (2024)
  18. A five-factor asset pricing model (2015) — link
  19. Absolving beta of volatility’s effects (2018) — link
  20. Simpler Better Market Betas (2019) — link
  21. Risk measurement when shares are subject to infrequent trading (1979) — link
  22. A NOTE ON USING CROSS-SECTIONAL INFORMATION IN BAYESIAN ESTIMATION OF SECURITY BETAS (1973) — link
  23. Trading Costs and Returns for U.S. Equities: Estimating Effective Costs from Daily Data (2009) — link
  24. A Simple Way to Estimate Bid-Ask Spreads from Daily High and Low Prices (2012) — link
  25. Market Microstructure Invariance: Empirical Hypotheses (2016) — link
  26. A Simple Estimation of Bid-Ask Spreads from Daily Close, High, and Low Prices (2017) — link
  27. A Taxonomy of Anomalies and Their Trading Costs (2015) — link
  28. A Transaction-Cost Perspective on the Multitude of Firm Characteristics (2020) — link

0 of 36 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.