A few days ago, I wrote about how although trading with the trend using simple price period look-backs is a simple and effective trading edge, the challenge remains as to how to pick the look back period. Traditionally, stock market methods have tended to use 12-month look-back periods, but the data shows that 6 months tends to work better, although 12 months remains profitable. In Forex, 3 months tends to work best, but recent years have shown it is only truly profitable with the major currency pairs. I explored three viable solutions to the problem of choosing a look back period:
A “best-of” look-back period, but then you have the problem of choosing a look-back period for the look-back period!
A dual look-back periods, with two periods both of which must be aligned. This is related to averaging those look-back periods which have tended to work well. I tend to use this in my own trading, where my favorite trades are in the three major currency pairs in the direction of both the 3-month and 6-month look-backs.
A relative volatility look-back, which is a new and interesting solution worth exploring. Here, you measure current volatility (range) against historical volatility (such as ATR). When volatility is higher than its average, you extend the look-back period, and vice versa when shorter. For example, if volatility is 150% of its average, and the baseline look-back period is 6 months, you would extend the look back period to 9 months (6 X 150%).
I had high hopes for the relative volatility look-back strategy but having tested this method on some major Forex pairs, I was disappointed to find that it was a total failure. I have seen evidence that it works on as part of a long-only momentum strategy on the S&P 500 Index, yet I feel that may be due more to the fact that the U.S. stock market tends to rise over time no matter how deep the pull-back. In my Forex tests I used baseline look-back periods of both 3 months and 6 months. The results for both are poor, so I am certain it is flawed as a concept.
My current conclusion is that the best approach to these problems is to find what has worked over a prolonged period and continue to apply it. I have no doubt that 3 months and 6 months seem to be the sweet spots.