Here's a 1996 study in which this was the case 1996 and 2003 follow-up, not within the past decade. The "significant" outperformance was that if the broader market rose 10.0%, the control group of splitters rose 10.8%. And an even smaller difference over three years. How much of the deviation could be explained by random variation from the sample size? Remember - it's quite easy to pick out, say 100 stocks with particular bullish variables, and find similar outperformance in hindsight. When you find such samples, even if you test and re-test and are able to replicate, the phenomena tends to go away over time (especially the past few years) as information sharing of such trends allows algorithms to price in such expectations more readily.
The bottom line: If you truly believe that old and small correlations (not causation) are statistically significant and will repeat, then just take a position in all splitting stocks, hedge verses broader market movement, and profit from the spread. Easy money, right? |
