Revisiting the Spark5
by David Forrest (firstname.lastname@example.org)
[Note: The following article is a reprint from last September that introduced one of our more popular screens from the Workshop Message Board. Check the end for links to some follow up info so that you can see how it has done since, and an article on still another favorite screen. -- Ann Coleman]
Alexandria, VA (April 23, 1999) -- Fools one and all, tonight's column is long overdue. For a long time now, we've read about mechanical models and backtesting. We started with Dow Dogs and relentlessly pursued better ways to beat "Beating the Dow." We've been influenced by O'Higgins, O'Shaughnessey, and O'Sheard. Now, it's time to look outward, to the community of Fools who have rallied around mechanical approaches and who constantly seek out even better ideas. Tonight is the equivalent of "open mike" night at the Fool and our spotlight shines on the Spark5.
For those who read the Workshop column on AOL, you may not be familiar with our friend Sparfarkle out on the website. We have two amazing groups of people contributing to the Workshop discussion, some on AOL, some on the web. While we can't bring web people into AOL to enjoy that discussion, I highly recommend that in addition to your AOL browsing that you visit out web Workshop board and see what's going on out there. Hopefully, in the future, we'll be able to merge the two and have one vibrant group of people.
Sparfarkle came onto the scene just about a month ago with the bold message board statement "This Strategy Beats UG5." When someone is claiming 14-year returns of 46% per annum, one's natural curiosity gets peaked, especially given the fact that this is the largest claimed return we've seen yet. Is it possible that this monster screen is for real? Let's take a look at the underpinnings.
I've read quite a few messages in this board from those who have complaints about rebalancing, commissions, lack of more backtesting, lack of data for annual rebalancing, and other problems with the Unemotional Growth strategy. Here's a similar strategy I've backtested all the way to 1984, so that results include the tough market of 1987. This is kind of similar to the UG strategy, and the core of the screen was not developed by me, but by James O'Shaughnessy in his book Invest Like The Best.
The criteria are as follows. Using Value Line, one sorts for:
1. 26-week price change greater than 20%
2. 1-yr EPS growth greater than 20%
3. EPS growth last quarter greater than 20%
4. Estimated EPS change 1 year out greater than 29%
5. Timeliness ranking of 1 or 2 (i.e. less than 3)
When you have this list, Sparfarkle recommends sorting the stocks by market capitalization (shares outstanding * price per share). Sparfarkle suggests buying the 5 largest-capitalized stocks and holding them for one year. If you'd like to take a look at some of the previous year's returns and the stocks associated with the backtesting, click the following links:
Current List as of August 5
Sparfarkle says the model has been tested since 1984, though only data for 1989 and higher has been presented. Hopefully, in the weeks and months to come, we can help backtest this even further and provide you all with the data. I'd also like to encourage everyone out there to test out their own models and report the results on the message boards. We'd like to continue spotlighting more and more of your ideas.
We'll talk again soon.
[Fools, you can see how the Spark5 is doing this year by clicking on 1999 Returns below, and you can get the latest stocks that qualify for this screen at New Rankings. And check out this Workshop Report from November for another winner: 50% a Year???]