High-frequency trading works for a few large firms. They have access to special computer systems and high-speed data connections that allow them to enter thousands of orders a minute.
The fastest firms can enter and exit trades in less than a second. Individuals cannot possibly compete with these firms and should focus on a longer time frame for their investments.
Hedge funds often hold their positions for a few days at a time. These firms have access to advanced trading algorithms and teams of analysts that evaluate information most individuals don't have access to. Information on industry trends -- like the impact of the Architecture Billings Index on the housing sector, for example, or the book-to-bill ratio for semiconductors -- can be valuable, but it's expensive and requires specialized knowledge to interpret.
After considering their edge, it's clear that individuals cannot really compete with hedge funds, either.
This means individuals may have the greatest level of success when they focus on holding positions for weeks or months. Several years ago, in my book "Smarter Investing in Any Economy," I demonstrated that the results from a weekly system are actually better than the results for a daily system.
Rather than setting a specific time frame, I believe a position should be held as long as it continues to outperform the market. Relative strength (RS) can be used to measure outperformance and can also be used as an indicator to sell.
Every week, I have been updating a system based on RS that uses the 26-week rate of change (ROC) to find the best ETFs to hold. This system beats the market, and since I started using it, the average holding period of a position has been 70 days. Because of the long holding period, I realized I could probably update this system less frequently.
Recently, I did some research on whether monthly trading could work. While the results were still good, they weren't as good. I then tested what would happen if I traded only once every two weeks and found there was a slight improvement compared with weekly trading. Performance suffered a small amount if trading was done only once every three weeks.
The results for a group of sector ETFs using the 26-week ROC system from January 1999 to the present are summarized below.
The size of losing trades decreased slightly as we slowed the frequency of trading. This is because the market tends to bounce back after a sharp drop, and the delay in trading captures part of that recovery.
Based on this research, I decided to start updating my 26-week ROC system every two weeks. I am confident that the results will not be hurt by this switch. Hopefully this change will make the system easier for readers to follow since there will be less trading required.
There are no changes in the 26-week ROC system at this time, and the model portfolio continues to hold: