Over the past months, as listed in the Drip archives, we have considered the types of companies that Fools might consider for their Drip portfolios. I have long felt that past performance should be a part of the criteria one considers, but have had no real evidence that a company's past performance has anything to do with its future.

We have all read that past performance is no guarantee of future results. As a matter of fact, this disclaimer is mentioned in more than 6,000 documents on the Web. However, I believe that there might indeed be a correlation between the two, so I began research to see if I could justify or debunk this statement.

When considering companies for a Drip portfolio, most investors tend toward the large companies. Since large is a relative term, I selected the nice round number of $100 billion for a market capitalization for my study. This gave me a list of the largest public companies today.

I then selected the companies from this group that have been the most successful over the past five years, in that their compound stock price growth has averaged more than 30%. This left me with companies that we all wish we bought five years ago, such as Microsoft (Nasdaq: MSFT), Intel (Nasdaq: INTC), Cisco Systems (Nasdaq: CSCO), Home Depot (NYSE: HD), Pfizer (NYSE: PFE), and others.

The question that I wanted to answer was how these companies compared to the S&P 500 Index over the previous five years, before their great outperformance. Of the 23 companies that had been selected in the screen, four did not exist as public companies 10 years ago, so they had to be removed from the mix.

I retrieved the monthly prices for the remaining 19 companies from July 1990 through June 1995, and adjusted the early price so that it could be compared to the S&P 500 Index. The result was that 17 of these companies had shown the ability to beat the index over the first half of the '90s. IBM (NYSE: IBM) and Nortel Networks (NYSE: NT) were the only two that failed to make the grade.

Unfortunately, this is only partial evidence to support that past performance has a correlation to future performance. It indeed tells us that today's largest and most successful companies would have compared successfully in long-term performance against the S&P 500 Index five years ago, and five years before that, too. However, there is a flaw in the study -- perhaps a fatal one.

The problem is that this study was done backward. We know what the successful companies are now, but I should have gone back five years and worked with the largest, most successful companies then, at that time.

I was able to find a resource that listed companies by market capitalization about five years ago. Examining the information from these companies five years back and five years ahead will definitely yield differing, though more reliable, results. However, the next problem becomes one of metrics.

I wonder how to define what would constitute a correlation. For instance, if one does not know anything about baseball and is told that a particular player fails six of every 10 times at bat, they might consider him to be a failure. On the other hand, most people know that the ability to bat .400 indicates a truly gifted athlete. (It's been nearly 60 years since anyone has reached this plateau, and only two have done so in the past 75 years.)

My question to Fools everywhere is: How do we learn what percentage of companies typically beats the S&P 500 Index over the course of any five years? I believe that, if this were known, we could compare this "benchmark" percentage against the number of large, successful companies five years ago that were able to beat the average. I am guessing that a very small minority of all companies accomplishes this feat regularly (as averages go), but I have no numbers to back up my thoughts.

What do you think? What should the metrics be to answer this question? Why not let us know on the Drip Investing Companies board?

I will come back to this subject two weeks from today, with the study of companies that were the biggest about five years ago, at which time we can look at the numbers.