Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
In a previous article I showed that the Coppock indicator could identify good buying opportunities for gold. This indicator has now been improved by making it more sensitive to the movements of the gold price which resulted in many more profitable buy signals. The last sell signal was at the end of November 2012 and no new buy signal has so far appeared. What is needed now for an early buy signal is a sustained upward move of the gold price to over $1,700 during the next few weeks, something which is certainly not outside the bounds of probability.
Figures 1, 2, 3 and 4 show the gold price, the Coppock indicator (blue graph), the buy signals (vertical black lines, 53 signals), and the model's gold investment periods (gray shaded periods). The percentage change in the year-over-year rolling returns for gold expressed in standard deviation terms for a 5-year rolling sample period is also shown (the yellow area graph, referred to as "yoy RRC% stdev").
Sell signals cannot directly be determined with this indicator, but a good strategy was to sell when the price had appreciated by 43%, but not later than one year (52 weeks) after a buy signal date. However, if a new buy signal appeared before the sell date applicable to the previous buy signal, then the sell criteria was applied from the new buy signal's date onwards, and so on. This had the effect of prolonging the gold holding periods if they contained numerous buy signals. The holding period and returns for each period are shown below, with only three of the twenty-two periods showing a small loss.
This model provided significantly better returns than the original model, as shown below. An investment in gold according to this model's signals produced a 14.91% compounded annual return from September 28, 1970 (the initial buy signal date) to November 23, 2012 (the last sell signal date), assuming no interest on cash when the model was not in gold. With interest at the Federal Funds rate from a money market account when the model was not in gold, the compounded annual return was 17.58%, equivalent to $100 growing to $92,785 over this period. The model's returns are considerably better than a continuous investment in gold which would have produced only a 9.63% average annual return over this period.
It would appear that the indicator is now in the early stages of forming a trough as shown in figure 2. Also, one can see that when the percentage change in the year-over-year rolling return of gold (expressed in standard deviation terms) was for an extended period of time significantly below zero, then this was usually followed by a buy signal and an upward move of the gold price. This is the case now and a buy signal should not be too far away. But when can we expect this signal to occur?
Figure 3 and 4 show forward projections of the weekly gold price to the end of 2013 which was determined with random numbers ranging from -$20 to +$25 for each weekly price change. This procedure generated a buy signal late in April, 2013 as shown in figure 3. If gold stays near the current level in the foreseeable future then no buy signal will appear, as shown in figure 4.
The weekly gold price was obtained from FRED for every Friday and Monday over the model's time span. The indicator was calculated using the Friday values. It's the sum of the 59-week rate of change and 49-week rate of change of the gold price, smoothed by a 37-week weighted moving average (WMA), multiplied by 10.
Indicator = 10 x WMA of (ROC + ROC)
The criteria for buy signals are:
- The level of the indicator minus the level of a 16-period exponential moving average of the indicator must be greater than 0.07,
- and the maximum level of the indicator over the preceding 39 weeks minus the current level of the indicator must be greater than or equal to 0.002.
Once a buy signal is generated (on a Friday) the investment is made the following Monday.
A sell signal is generated when:
- the gold price has appreciated for the first time by 43% or more from the buy price, as determined from the Monday gold prices, but not later than 52 weeks after the buy date.
However, should a new buy signal appear before the sell signal applicable to the preceding buy signal has occurred, then the sell criteria apply renewed from the new buy signal onwards, and so on.
The refinements to this model were made in collaboration with Don Patton.
Georg Vrba is a professional engineer who has been a consulting engineer for many years. In his opinion, mathematical models provide better guidance to market direction than financial "experts." He has developed financial models for the stock market, the bond market and the yield curve, all published in Advisor Perspectives. The models are updated weekly. If you are interested to receive theses updates at no cost, send email request to firstname.lastname@example.org.