About a week ago, the Wall Street Journal ran an article about the Value Line ranking system for stocks and its inventor, statistician Sam Eisenstadt.
I knew Sam in the late 1970s – early 1980s, during the heyday of Value Line. At that time, the company was an incubator that launched the careers of a large number of successful investors. The ranking system was also a formula for consistent outperformance.
Then the music began to stop.
What I find most interesting about the WSJ article is Sam’s belief that the Value Line ranking system will begin to work again.
how the VL ranking system works
The method, which was revolutionary when it was introduced in the middle of the last century, is taken straight out of a finance textbook.
It evaluates stocks by analyzing two main variables, cheapness and growth:
Cheapness is measured by taking the current price-earnings ratio for each stock and seeing where it stands relative to its PE over the prior ten years. If the current PE is the lowest, the stock receives the highest score. If the current PE is the highest, the stock gets the lowest score.
Growth is measured by taking the current per-share earnings growth rate and comparing it with the company’s earnings growth rate in each of the past ten years. If the rate of growth is currently the highest, the stock gets the highest score. If the growth rate is the currently the lowest, the stock gets the lowest score.
After scores for each stock are tallied, the totals are compared with those of all the other stocks in the VL universe of about 1,800 stocks–in the early days, this required a mainframe. A couple of technical variables are mixed in. The factors are weighted (this is the system’s secret sauce). The end result is a ranking of the universe in order from 1 to 1,800.
The stocks are then grouped on a bell curve:
–the top 100 are ranked 1
–the next 300 are ranked 2
–the middle thousand are ranked 3
–the next 300 are ranked 4
–the bottom 100 are ranked 5.
Fresh rankings are published each week.
results
For over twenty years, the system worked like a charm. 1s consistently outperformed the market; 5s underperformed (in fact, academic research showed that 5s underperformed more deeply and for longer periods than 1s outperformed). In most years, stocks performed precisely in line with their ranks. That is, 1 outperformed 2s, which outperformed 3s, which outperformed 4s, which outperformed 5s.
Academics were flummoxed. The system was statistically sound. It used only publicly available historical data. And yet, contrary to the “efficient markets hypothesis” (the academic assumption that, in simple terms, all publicly available information is immediately factored into stock prices), favorably ranked Value Line stocks outperformed as a group year after year after year.
Then, suddenly, the system didn’t work so well. The WSJ article has a chart that shows the ten-year annualized performance of the VL 1s minus the performance of the 5s. That outperformance figure peaks during the first half of the 1980s at a staggering 40% difference per year over the prior decade. It then begins to fall pretty steadily through 2006, when the performance difference for the prior decade is close to zero.
(An aside: one might question whether a 10-year time frame isn’t a bit too long or whether having the top 5% or so do better than the absolute bottom of the barrel is a high enough bar–but the author of the article, Mark Hulbert, didn’t go there and I won’t either.)
What happened? Are the bad times over? That’s for tomorrow’s post.