Researcher course in systematic investing
Fama-French Factor Models
The empirical asset pricing toolkit: the Fama-French three- and five-factor models, time-series regressions and the GRS test, Fama-MacBeth cross-sectional regressions, standard errors, and how to judge whether a strategy has alpha.
In 1993 Eugene Fama and Kenneth French proposed what is now called the Fama-French model: that two factors, size and value, together with the market, explain most of the differences in average stock returns that the CAPM could not. Their models and methods became the standard way to test whether an investment strategy earns returns beyond known risk factors.
How do you test whether a strategy has alpha?
Regress its returns on the factor returns. The intercept is the return the factors do not explain, and the coefficients show the strategy’s exposures. A strategy that looks profitable often turns out to be a mix of known factors. The course shows how to run and interpret that test, and how to test many portfolios jointly with the Gibbons, Ross and Shanken test.
What is a Fama-MacBeth regression?
A two-step method for testing whether a characteristic is priced across stocks: run a cross-sectional regression every month, then average the coefficients over time. It remains one of the most used methods in empirical finance. You will implement it, with corrections for the errors it can introduce.
Who is this course for?
Researchers, graduate students and quant analysts who need the tools of empirical asset pricing. It counts toward the Certificate in Systematic Investing.
What you will be able to do
- Explain the construction of the Fama-French factors and replicate them.
- Run time-series factor regressions and interpret alpha and loadings.
- Apply the GRS test to a set of portfolios.
- Run Fama-MacBeth regressions with appropriate standard errors.
- Compare competing factor models and judge their evidence.
Syllabus, module by module
From CAPM to factors
The anomalies that the CAPM could not explain, and the response of Fama and French in 1993.
Building the factors
Size and book-to-market sorts, SMB and HML, and replicating them from raw data.
Time-series tests
Factor regressions, alpha, the GRS test and what a rejected model means.
Cross-sectional tests
Fama-MacBeth regressions, errors-in-variables and Shanken corrections.
The five-factor model
Profitability and investment factors, the fate of HML, and momentum's absence.
Comparing models
Carhart, q-factor and other models, and how to choose between them.
Assessment preparation
Supervised work on the written assessment.
How the certificate is earned
A written paper-style assessment: you test a strategy or anomaly against several factor models, with time-series and cross-sectional methods, and write up the results as a short paper. On a pass you receive the Course certificate in Fama-French Factor Models, issued by the Institute with a verification code. Read how certification works.
Before you start
Regression and hypothesis testing, working Python, and the material of Factor Investing.
This course counts toward
Questions applicants ask
What are the Fama-French factors?
The three-factor model adds two factors to the market: SMB, small minus big, for company size, and HML, high minus low, for book-to-market value. The 2015 five-factor model adds RMW for profitability and CMA for investment.
Where does the factor data come from?
Kenneth French publishes factor returns on his website, and the course uses them. You will also build simplified versions from raw data to see how the choices in construction affect results.