In the first part of this series we discussed the impracticality of inverse cap weighting (also note Michael James’ comments at the bottom of that post.) In this part we will look at what needs to be done to sever the link between price and weight so as to avoid overweighting overvalued stocks and underweighting undervalued stocks.
Not intending to be patronizing: to sever the link between price and weight you must weight stocks based on something not directly related to price. The question then, is to figure out what metric to use. Many people use stock market indices as long term proxies for measuring the growth of an economy. If that is a loose assumption we can make, then any metric which accomplishes this can be used. We could look at the number of employees in a company versus all employees in all companies. We could look at the number of telephone lines in a company versus all the telephone lines in all companies. I’ll see if I can dig up the actual research, but these two methodologies both outperformed cap-weighting over long periods of time.
Whether one uses these metrics (they don’t in practice) or metrics found in the audited financial statements (like gross sales), they all accomplish the goal of randomizing pricing errors instead of magnifying the overpriced weightings and decreasing the weights to underpriced companies. This is the important concept in all of this and it is spelled out in detail in this post.
Disclosure: I work for a company which manufactures index funds which seek to break the link between portfolio weight and stock price, so take all of this with the grain of salt that I have a bias.