famafrench - Package Documentation¶
Toolbox for constructing and replicating datasets from Ken French’s online data library by accessing WRDS remotely through its cloud server.
famafrench’s current efficient performance results from features such as the use of a least recently used (LRU) cache implemented using Python’s functools.lru_cache().
Future plans are to continue to expand the toolbox to include additional factor-based datasets relevant for empirical asset pricing. These include the following:
-
Betting Against Beta (BAB)
Quality Minus Junk (QMJ)
Modified Value - High Minus Low (HMLD)
Lettau, Ludvigson, and Ma (2019) Capital Share Factor:
Capital Share of Aggregate Income (KS)
Pastor and Stambaugh (2003) Liquidity Factors:
Non-Traded Liquidity Factor
Traded Liquidity Factor
Sadka (2006) Liquidity Factors:
Fixed-Transitory Factor
Variable-Permanent Factor
Stambaugh and Yuan (2017) Clustered Mispricing Factors:
Management-related Factor (MGMT)
Performance-related Factor (PERF)
Mispricing (non-clustered) Factor (UMO)
Please reach out if you have any recommendations or suggestions for improvements. Collaborations are welcomed - reach out at chris.jauregui@berkeley.edu!
Contents¶
What’s New?!
Getting Started
Applications and Examples
Connecting to the WRDS cloud server
WRDS Query Tools
Estimating Market Betas and Rolling Residual Variances
Constructing Portfolios and Return-Based Factors
Comparing to Ken French’s Online Library
Summary Statistics and Diagnostics
Auxiliary Functions and Utilities
API Reference
Change Log
How to Cite¶
This package (and its release as of April 20, 2020) should be cited using fishare. For example, for the 0.1.0 release,
[*] Jauregui, Christian (2020): famafrench Python library. Release 0.1.0 (Version 0.1.0). figshare. Software. https://doi.org/10.6084/m9.figshare.12170439
Todo¶
Todo
Ken French’s data library documentation notes the following regarding the construction of their daily portfolio returns:
In May 2015, we made two changes in the way we compute daily portfolio returns so the process is closer to the way we compute monthly portfolio returns. In daily files produced in May 2015 or thereafter, stocks are dropped from a portfolio immediately after their CRSP delist date; in files produced before May 2015, those stocks are held until the portfolio is reconstituted, at the end of June. Also, in daily files produced before May 2015 we exclude a stock from portfolios during any period in which it is missing prices for more than 10 consecutive trading no price for more than 200 consecutive trading days.
Future versions will verify the aforementioned adjustments are accounted for in the package’s construction of daily portfolio returns.
Todo
Include option to construct the Fama and French (2018) cash-based profitability factor,
RMWc.Within the instance method used for constructing market betas and rolling residual variances,
getFactorResults(), extend the Dimson (1979) methodology based on Scholes and Williams (1977) to other factor quantities of risk beyond the market (CAPM) beta (eg, SMB and HML quantities of risk).Verify the pandas-datareader Python library is still unable to pull monthly and annual datafiles for the Short-Term Reversal or Long-Term Reversal Fama-French-style factors made public through Ken French’s online library.
When applicable, constructed portfolios, specifically portfolio returns, number of firms in each portfolio, and average anomaly portfolio characteristics are compared with those provided by Ken French for the same frequency and over the same period. The sample Pearson correlations, sample means, and sample standard deviations for the following portfolios can be improved:
To improve the statistical metrics, adjustments in how the aforementioned anomaly characteristics are computed or estimated will be incorporated in future releases.