According to the academic study "Underreaction to Self-Selected News Events: The Case of Stock Splits", which looked at stocks that declared splits between 1988 and 1997, the mean return of stocks in this group was 9% higher in the first year after the split and the median return was 6.37% higher when compared to a control group.

An earlier paper called "What Do Stock Splits Really Signal?" by professors David Ikenberry, Graeme Rankine and Earl K. Stice looked at the returns of 1,275 companies that declared stock splits between 1975 and 1990 and found that in the first year after the effective date of the split, these stocks outperformed a benchmark group of companies by 8%. Over a period of three years, the results were even more impressive with these stocks outperforming the benchmark group by 16%.

So it looks like there may be plenty of time to get into companies declaring stock splits. In my opinion, the best time may be right after the announcement. It also depends on general market conditions, the nature of the industry and if the announcement causes a large spike in the stock price.  

Many investors consider stock splits as insignificant events. If that perception changes and this stock split theory become as popular as the Dogs of the Dow, the returns discussed in these academic papers may actually diminish.   

But the geek question to ask is:

Do they outperform at 9% because of the split, or is it that they split because they are a good company, and expected to outperform?

Take BRKA  which has never split but has outperformed the dow over the last tw as the ‘control’ and what would happen?

http://finance.yahoo.com/q/bc?s=BRK-A&t=2y&l=on&z=m&q=l&c=%5EDJI

"The reason for these higher returns? After ruling out factors such as increased coverage by analysts, Ikenberry and Ramnath concluded that 85% of companies that declare stock splits tend to have higher earnings growth in the ensuing years and analysts that follow these companies are slow to revise their earnings expectations upwards."

A pure stock split strategy may not always work, especially during a raging bull market when many companies split their stock simply because a rising tide lifts all boats.

Using Berkshire Hathaway (BRK-A) as a control group is an excellent idea. I think that it depends entirely on the company that declares a stock split and the period of time you are looking at. For example, while BRK has outperformed the Dow over the last two years, check out this chart comparing the performance of the Dow, Berkshire Hathaway and Logitech.

I am not sure if there are any academic studies done on the effect of reverse splits but if I find anything, I will post it on this thread.