I recently wrote about how the production of butter in Bangladesh was demonstrated to be an effective predictor of the U.S. stock market's future. I had run across that valuable piece of information -- stemming from the research of Cal Tech professor David Leinweber -- in the latest issue of our Motley Fool Stock Advisor newsletter service.

Well, as sometimes happens, I soon heard from the horse's mouth. Professor Leinweber emailed me a copy of a more recent and more detailed study he'd undertaken, along similar lines. This one was titled "Stupid Data Miner Tricks." This time around, Leinweber had found much more accurate market predictors. His three variables, in fact, accounted for fully 99% of the S&P 500's performance over 10 whole years. The three variables? Butter production in Bangladesh, U.S. and Bangladeshi cheese production, and the population of sheep in the U.S. and Bangladesh.

I'll pause now, so you can stop laughing and can concentrate again on this important matter. It's funny, sure. But if you were letting these three factors guide your investments between 1983 and 1993, they seriously would have been able to make you very rich.

Here's the problem, though. Outside that time period, the guidance is -- and here you can choose your favorite words from the professor -- "utterly useless," "a total crock," or "just a chance association which would inevitably show up if you look at enough data series, as we did."

The danger, as the good professor illustrates, is that we sometimes really do get swayed by results like these -- because they may not always involve silly things like butter or sheep. The results might seem much more convincing and reliable, for instance, if they're tied to various global economies, to the housing market, or to interest rates. Yet if these findings are the result of data mining -- digging into history to find meaningful past patterns -- then they, too, may also be a total crock.

Looking forward
For more forward-looking market predictions, I invite you to explore our new world of Motley Fool CAPS, a community of tens of thousands of investors predicting the movement of thousands of stocks. Instead of relying on backtesting, you can find some CAPS participants with very strong track records (we highlight these folks for you), and you can then follow their new predictions. Some highly rated CAPS stocks include Suntech Power (NYSE:STP), PetroChina (NYSE:PTR), Brookfield Asset Management (NYSE:BAM), and McDermott International (NYSE:MDR).

Joining CAPS is free. And it will surely serve you better than counting sheep.

Longtime Fool contributor Selena Maranjian does not own shares of any companies mentioned in this article. Brookfield Asset Management is a Motley Fool Global Gains recommendation. Suntech Power is a Motley Fool Rule Breakers recommendation. Try any of our investing services free for 30 days. The Motley Fool is Fools writing for Fools.