getfactormodels 0.0.7

Download datasets for various multifactor asset pricing models.

getfactormodels

Reliably retrieve data for various multi-factor asset pricing models.

Models

• The 3-factor, 5-factor, and 6-factor models of Fama & French ^{[1] [3] [4]}
• Mark Carhart's 4-factor model ^[2]
• Pastor and Stambaugh's liquidity factors ^[5]
• Mispricing factors of Stambaugh and Yuan^[6]
• The $q$-factor model of Hou, Mo, Xue and Zhang^[7]
• The augmented $q^5$-factor model of Hou, Mo, Xue and Zhang^[8]
• Intermediary Capital Ratio (ICR) of He, Kelly & Manela^[9]
• The DHS behavioural factors of Daniel, Hirshleifer & Sun^[10]
• The HML $^{DEVIL}$ factor of Asness & Frazzini^[11]
• The 6-factor model of Barillas and Shanken^[12]

Thanks to: Kenneth French, Robert Stambaugh, Lin Sun, Zhiguo He, AQR Capital Management (AQR.com) and Hou, Xue and Zhang (global-q.org), for their research and for the datasets they provide.

Installation

[!IMPORTANT]

getfactormodels is pre-alpha (until version 0.1.0), don't rely on it for anything. Requires:

• Python: >=3.10

pip install -U getfactormodels

Quick start

Basic usage:

• Import getfactormodels and use the get_factors function, with a model parameter:

 import getfactormodels

 data = getfactormodels.get_factors(model='q', frequency='d')

• All other parameters are optional. By default monthly data is returned.

import getfactormodels as gfm

# Download Fama-French 3-factor daily data since start_date:
df = gfm.get_factors(model='ff3', frequency='d', start_date='2006-01-01')

# Monthly DHS factors until end_date:
dhs = gfm.get_models(model='dhs', end_date='2010-12-31'

# Mispricing factors (monthly) with date range and export:
df = gfm.get_factors(
    model='mispricing',
    start_date='1970-01-01',
    end_date='1999-12-31',
    output='mispricing_factors.csv'  #.csv, .txt, .pkl
)

• You can import only the models you need. For example, the ICR and q-factor models:

from getfactormodels import ICRFactors, QFactors

# Passing a model class without params defaults to monthly data.
icr = ICRFactors() # look! no params!

# Use the download module to get the data
df = icr.download()

# The 'q' models, and the 3-factor model of Fama-French have weekly data available
df = QFactors(frequency='w',
            start_date='1992-05-22',
            end_date='2019-01-05').download() # chained! Wow!

#print(df.tail(3))

• All model Classes: HMLDevil, CarhartFactors, FamaFrenchFactors, QFactors, LiquidityFactors, MispricingFactors, BarillasShankenFactors, ICRFactors, DHSFactors.

Parameters

• All model classes support:
  • frequency: d w m y
  • start_date: YYYY-[MM-DD] format
  • end_date: YYYY-[MM-DD] format
  • output_file: Export path

• FamaFrenchFactors has a model param. Accepts 3 4 5 6 or equivilent model name (ff3, famafrench3) (default: 3).
• QFactors has a classic boolean param (default: false) for returing the 4-factor q-factor model (2015).

CLI

Requires: bash >=4.2

This is old but should still work until redo.

• You can also use getfactormodels from the command line. It's very basic at the moment, here's the --help:

  $ getfactormodels -h
  
  usage: getfactormodels [-h] -m MODEL [-f FREQ] [-s START] [-e END] [-o OUTPUT] [--no_rf] [--no_mkt]
  

• An example of how to use the CLI to retrieve the Fama-French 3-factor model data:

  $ getfactormodels --model ff3 --frequency M --start-date 1960-01-01 --end-date 2020-12-31 --output .csv
  

• Here's another example that retrieves the annual 5-factor data of Fama-French, without the RF column (using --no[_]rf)

  $ getfactormodels -m ff5 -f Y -s 1960-01-01 -e 2020-12-31 --norf -o ~/some_dir/filename.xlsx
  

• To return the factors without the risk-free rate RF, or the excess market return Mkt-RF, columns:

  $ getfactormodels -m ff5 -f Y -s 1960-01-01 -e 2020-12-31 --norf --nomkt -o ~/some_dir/filename.xlsx
  

Data Availability

This table shows each model's start date, available frequencies, and the latest datapoint if not current. The id column contains the shortest identifier for each model. These should all work in python and the CLI.

id	Factor Model	Start	D	W	M	Q	Y	End
3	Fama-French 3	1926-07-01	$\checkmark$	$\checkmark$	$\checkmark$		$\checkmark$	-
4	Carhart 4	1926-11-03	$\checkmark$		$\checkmark$		$\checkmark$	-
5	Fama-French 5	1963-07-01	$\checkmark$		$\checkmark$		$\checkmark$	-
6	Fama-French 6	1963-07-01	$\checkmark$		$\checkmark$		$\checkmark$	-
hmld	HML $^{DEVIL}$	1990-07-02	$\checkmark$		$\checkmark$			-
dhs	DHS	1972-07-03	$\checkmark$		$\checkmark$			2023-12-29
icr	ICR	1970-01-31 Daily: 1999-05-03	$\checkmark$		$\checkmark$	$\checkmark$		2025-06-27
mis	Mispricing	1963-01-02	$\checkmark$		$\checkmark$			2016-12-30
liq	Liquidity	1962-08-31			$\checkmark$			2024-12-31
q q4	$q^5$-factors $q$-factors	1967-01-03	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	2022-12-30
bs	Barillas-Shanken	1967-01-03	$\checkmark$		$\checkmark$			2024-12-31

[TODO] Docs!

References

Publications:

1. E. F. Fama and K. R. French, ‘Common risk factors in the returns on stocks and bonds’, Journal of Financial Economics, vol. 33, no. 1, pp. 3–56, 1993. PDF
2. M. Carhart, ‘On Persistence in Mutual Fund Performance’, Journal of Finance, vol. 52, no. 1, pp. 57–82, 1997. PDF
3. E. F. Fama and K. R. French, ‘A five-factor asset pricing model’, Journal of Financial Economics, vol. 116, no. 1, pp. 1–22, 2015. PDF
4. E. F. Fama and K. R. French, ‘Choosing factors’, Journal of Financial Economics, vol. 128, no. 2, pp. 234–252, 2018. PDF
5. L. Pastor and R. Stambaugh, ‘Liquidity Risk and Expected Stock Returns’, Journal of Political Economy, vol. 111, no. 3, pp. 642–685, 2003. PDF
6. R. F. Stambaugh and Y. Yuan, ‘Mispricing Factors’, The Review of Financial Studies, vol. 30, no. 4, pp. 1270–1315, 12 2016. PDF
7. K. Hou, H. Mo, C. Xue, and L. Zhang, ‘Which Factors?’, National Bureau of Economic Research, Inc, 2014. PDF
8. K. Hou, H. Mo, C. Xue, and L. Zhang, ‘An Augmented q-Factor Model with Expected Growth*’, Review of Finance, vol. 25, no. 1, pp. 1–41, 02 2020. PDF
9. Z. He, B. Kelly, and A. Manela, ‘Intermediary asset pricing: New evidence from many asset classes’, Journal of Financial Economics, vol. 126, no. 1, pp. 1–35, 2017. PDF
10. K. Daniel, D. Hirshleifer, and L. Sun, ‘Short- and Long-Horizon Behavioral Factors’, Review of Financial Studies, vol. 33, no. 4, pp. 1673–1736, 2020. PDF
11. C. Asness and A. Frazzini, ‘The Devil in HML’s Details’, The Journal of Portfolio Management, vol. 39, pp. 49–68, 2013. PDF
12. F. Barillas and J. Shanken, ‘Comparing Asset Pricing Models’, Journal of Finance, vol. 73, no. 2, pp. 715–754, 2018. PDF

Data sources:

• K. French, "Data Library," Tuck School of Business at Dartmouth. Link
• R. Stambaugh, "Liquidity" and "Mispricing" factor datasets, Wharton School, University of Pennsylvania. Link
• Z. He, "Intermediary Capital Ratio and Risk Factor" dataset, zhiguohe.net. Link
• K. Hou, G. Xue, R. Zhang, "The Hou-Xue-Zhang q-factors data library," at global-q.org. Link
• AQR Capital Management's Data Sets.
• Lin Sun, DHS Behavioural factors Link

(back to top)

License

Known issues

• The first hml_devil_factors() retrieval is slow, because the download from aqr.com is slow. It's the only model implementing a cache—daily data expires at the end of the day, and will only re-download when the requested end_date exceeds the cache's latest index date. Similar for monthly, expiring at at the end of the month, and re-downloaded when next needed.
• Some models aren't downloading. Update: all models should be downloading.

Todo

• Refactor: a complete rewrite, implementing a better interface and design patterns, dropping dependencies.
• Docs
• Every model should have an about and author/copyright info, and general disclaimer
• This README
  • Examples
• Tests
• Error handling

IGNORE THIS

There's also the FactorExtractor class (which doesn't do much yet, it's mainly used by the CLI):

from getfactormodels import FactorExtractor

fe = FactorExtractor(model='carhart', start_date='1980-01-01', end_date='1980-05-01)
fe.get_factors()
fe.drop_rf() 
fe.to_file('~/carhart_factors.csv')

.drop_rf() will return the DataFrame without the RF column. You can also drop the Mkt-RF column with .drop_mkt()

REWRITE ME