Q. You became a mutual fund timer long before it became popular. What was your original inspiration?
A. Well, I really owe it to a friend. I remember the day as if it were yesterday. I wandered into a colleague’s office, and he said, Hey, Gil, take a look at these numbers. He had invested in a municipal bond fund to take advantage of the prevailing high interest rates, which at the time were about 10 to 11% tax free. Although he was getting a high interest rate, he discovered that his total return was actually declining rapidly because of the steady attrition in the net asset value (NAV). He handed me a sheet with about a month’s worth of numbers, and I noticed that the trend was very persistent: the NAV had declined for approximately 22 consecutive days. He said, Fidelity allows you to switch into a cash fund at any time at no charge. Why couldn’t I just switch into a cash fund at any time at no charge. Why couldn’t I just switch out of the fund into cash when it started to go down and then switch back into the fund when it started to go back up? My reactions was, Nick, I don’t think the markets work that way. Have you ever read A Random Walk Down Wall Street? I pooh-poohed his idea. I said, The problem is that you don’t have enough data. Get some more data, and I bet that you’ll find this is not something you could make any money on over the long run. He did get more data, and, amazingly, the persistency of trends seemed to hold up. I quickly became convinced that there was definitely something nonrandom about the behavior of municipal bond funds. It was the simplest approached that proved the best. We called it the one penny rule. In the two year’s worth of date we obtained, we found that there was approximately an 83% probability that any uptick or downtick day would be followed by a day with a price move in the same direction. In the spring of 1980, I began to trade Fidelity’s municipal bond fund in my own account based on this observation.

Trend Following Products

Review trend following systems and training:

More info here.