A reader observed and suggested:
“When I first started paying attention to markets in the 1980s and 1990s, one frequently cited argument against market efficiency was the Value Line anomaly – the fact that stocks with their best timeliness ranking had extraordinary returns over a long period. You can still find charts showing how well Group 1 has done versus Group 5 over a multi-decade period, but it seems that there has not been much cumulative performance separation among groups in recent years. Some raw data on their site shows that the predictive power of the ranking system seems to be missing from about 2000 onward. It might be interesting to look at what was once a widely discussed method of potential market outperformance.”
The Value Line Timeliness Ranking System sorts stocks into five groups, with Group 1 (5) expected to exhibit the strongest (weakest) future performance. Value Line summarizes annual performance data for Groups 1 through 5 based on assumptions of both weekly and annual group re-sorting. Because the trading frictions of weekly re-sorting are likely high and difficult to estimate, we focus on performance by group for annual re-sorting. Specifically, we measure the Group 1 annual returns minus the Group 5 annual returns and the Group 2 annual returns minus the Group 4 annual returns. If the ranking system is persistently reliable, both sets of differences should be persistently positive, with the differences for the first set generally larger than those for the second set. Using annual return data stated by Value Line for 1965 (partial year) through 2008 (nearly 44 years), we find that:
The following chart plots the Group 1 annual returns minus the Group 5 annual returns based on annual re-sorting over the entire sample period, along with a best-fit linear trend line. The trend line indicates that the differences in annual returns between Group 1 and Group 5 have roughly disappeared over the sample period. While the results for the early 2000s strongly influence the trend line, the slope began to turn negative in the early 1990s. The average for the series over the last decade is negative.
The next chart plots the Group 2 annual returns minus the Group 4 annual returns based on annual re-sorting over the entire sample period, along with a best-fit linear trend line. As would be expected if the ranking system works, this trend line is mostly lower than that for Group 1 minus Group 5. However, it too indicates that the differences in annual returns between the higher-ranked group and the lower-ranked group have roughly disappeared over the sample period. The slope of this trend line also began to turn negative in the early 1990s. The average for this series over the last decade is negative.
Has Value Line hit a streak of bad luck? Has nimble competition made exploiting their valuation system a lot harder? Has the era of computerized quant trading changed the rules for stock valuation? Perhaps the trend reflects some combination of these factors.
The sample size of 44 years is not large for the purpose of determining annual return trends, such that a few additional years of unusual performance could substantially shift the trend lines. Incorporating the trading frictions of annual re-sorting would lower the return differences depicted above.
In summary, limited evidence indicates that any abnormal performance of the the Value Line Timeliness Ranking System may have disappeared.
