Value Investing GitHub Repository
We are fascinated by the empirical evidence showing that value investing has outperformed the market most of the time for as far back in history as the data allows us to see. It fascinates us because it seems counterintuitive that a well-known, simple process could survive for long without becoming so popular that it no longer worked.
There is hot debate about the cause of this anomaly. One school of thought is that value stocks are riskier than the market as a whole and investors are compensated with higher expected returns for the additional risk. Another school of thought is that human behavior — our animal spirits — causes investors to be irrational, which, in turn, sometimes causes a stock to be priced lower (or higher) than what the underlying company’s inherent worth would justify. A third school of thought is that maybe a little of both phenomena, risk compensation and our animal spirits, are at play.
Regardless of the cause, to help illustrate the empirical evidence of this phenomenon, we put together a GitHub repository of scripts that allows anyone to run one of several data analysis experiments related to the value anomaly. It is organized as a series of “studies”, and each one demonstrates an empirical aspect of value investing. To execute a study, all a user needs is an up-to-date installation of git and the statistical programming language “R”.
Many of the studies were not originally conceived by Euclidean. In particular, William Bernstein first published a version of the value-and-inflation study in 2001. Star Capital published a version of the value-vs-growth analysis in 2016. However, this repository is meant to be a self-contained, one-stop-shop for reproducing these studies and others from the original source data. We also hope that it will serve as an easy-to-use launching point for future studies on this topic.
One does not need to have access to proprietary data to execute these studies. Rather, each script downloads the required data necessary to execute them from the original source data. We would like to thank Kenneth R. French and Robert Shiller for making this data available through their websites here and here, respectively. The equity investment return data is copyrighted by Kenneth R. French.
Finally, it is important to point out that, in these studies, value stocks are defined as companies with high book equity to market value (book-to-market). This is a very crude way of defining a value stock. Examining only the result of a company’s assets minus its liabilities (its book value) is akin to looking at a company as if it was merely a bank account. Yet, there are so many other factors that comprise a company’s intrinsic value, such as its products, customers, brands, operating capacity, and intellectual property. Perhaps this is why many researchers have found that other measures of value yield better results than book-to-market when building value stock portfolios.
This said, the book-to-market time series that Ken French provides on his website offers a unique opportunity. No other data source goes as far back in time or is freely available for public consumption. Therefore, we believe our repository is useful for examining the high-level behavior of value investing and how it has performed relative to other investment approaches over long periods of time.