Never Buy Expensive Stocks. Period. by Wesley R. Gray, Ph.D.
We did a recent internal simulation study on the performance of cheap and expensive stocks based on a variety of valuation metrics.
We looked at all our favorites from our JPM paper “Analyzing Valuation Measures: A Performance Horse Race over the Past 40 Years:”
- EBIT/TEV
- EBITDA/TEV
- B/M
- Gross Profits / TEV
- FCF / TEV
This research is part of a larger academic paper, but I did want to highlight one aspect of our study that we thought more practitioner-minded readers would find fascinating.
Note: We focus our results on EBIT/TEV because it is our preferred measure for identifying “cheapness.” The results for other valuation metrics are quantitatively similar.
How Does Our Simulation Work?
- First, break stocks down into different valuation deciles from 1963 to 2013 based on EBIT/TEV (we only focus on US mid/large cap to avoid weird micro/small cap outlier effects).
For example, if there are 1000 stocks, stocks 1-100 go in the first decile; stocks 101-200 go in the second decile, etc.
- Next, do 1000 simulations of random 30 stock portfolios drawn from the cheap stock decile or the expensive stock decile.
For example, simulation #1 draws 30 random stocks each month from the top and bottom decile from 1963 to 2013. This is the rough equivalent of saying, “we are going to have a monkey throw 30 darts,” every month during the 50 year period, to establish in each month separate 30 stock portfolios. Once our monkey has thrown his 30 darts in each month, we will then have 600 separate monthly portfolios (12 months * 50 years) and will have made 18,000 (30 stocks * 600 months) individual stock picks. This represents one simulation. We do 1000 simulations for the top decile and 1000 simulations for the lowest decile.
- Calculate the performance statistics for each simulated strategy from 1963 to 2013.
Each simulated strategy represents the returns a value-investing monkey (cheap stock buyer) or growth-investing monkey (expensive stock buyer) would achieve over the full time period. We calculate compound annual growth rates (CAGR), standard deviation, and maximum drawdown.
- Tabulate the performance statistics for all 1000 simulations.
We compile the results in the charts/tables below.
Full article via Alpha Architect
