It has been about one-and-half to two weeks since I discussed the the All that Glitters models. Frankly, I have been consumed with other activities. Additionally, the models did not change significantly over the preceding weeks and I did not want to sound like a broken record- I know, too late. Lastly, I have continued to ponder various risk controls measures in the model and I would like to report on the promising front runner.
First, the models. The longer dated calculations remains in strong buy regions with the 1-year model sitting at -2 and the 2-year model calculation coming in at -2.5. That said, the 6-month model calculation has turned away from the -2 demarcation in recent weeks due to the recent increase in the price of gold now approaching $1,300 per ounce. The models are presented below.
6-Month Model, -1.4
1-Year Model, -2
2-Year Model, -2.5
As stated above, I have continued to look at numerous measures of risk control, as I wanted to avoid future situations like the one seen in the last few months, having lost money on gold/precious metal investments. As I spun towards various levels of short-term regret and reexamined the models, It dawned on me that maybe the risk control I was seeking did not require more data and/or additional data. In most instances, the absolute level of the model results do a fantastic job at identifying potential bottoms or tops in gold/precious metal shares. However and in many other instances, the change in the trend of models is equally important.
By itself, any trend or slope coefficient analysis of the models produce inferior performance results relative to either straight model performance or more importantly the buy-and-hold case. Combine a slope analysis with the model, however, the overall performance improves. In some circumstances, significantly so. For instance, the following table shows the performance results of the Philly Gold/Silver Index for 6-month periods after the 1-year model falls below either -1 or -2 demarcation. The performance is shown with and without the slope modification with a slope modification defined as a positive slope observed on the 6-month model.
It appears that adding a slope modification vastly reduces the risk of a significant loss in the historical experience. This is as the minimum downside is vastly reduced with the slope modification.
I am continuing to test various time periods and base model used in the slope calculation but this risk control seems promising. Presently, the 6-month slope is positive and in conjunction with the models suggest that positions in gold/precious metal shares should be accumulated. How does this differ from my call a couple months ago. For starters, the 6-month slope was negative then.
First, the models. The longer dated calculations remains in strong buy regions with the 1-year model sitting at -2 and the 2-year model calculation coming in at -2.5. That said, the 6-month model calculation has turned away from the -2 demarcation in recent weeks due to the recent increase in the price of gold now approaching $1,300 per ounce. The models are presented below.
6-Month Model, -1.4
1-Year Model, -2
2-Year Model, -2.5
As stated above, I have continued to look at numerous measures of risk control, as I wanted to avoid future situations like the one seen in the last few months, having lost money on gold/precious metal investments. As I spun towards various levels of short-term regret and reexamined the models, It dawned on me that maybe the risk control I was seeking did not require more data and/or additional data. In most instances, the absolute level of the model results do a fantastic job at identifying potential bottoms or tops in gold/precious metal shares. However and in many other instances, the change in the trend of models is equally important.
By itself, any trend or slope coefficient analysis of the models produce inferior performance results relative to either straight model performance or more importantly the buy-and-hold case. Combine a slope analysis with the model, however, the overall performance improves. In some circumstances, significantly so. For instance, the following table shows the performance results of the Philly Gold/Silver Index for 6-month periods after the 1-year model falls below either -1 or -2 demarcation. The performance is shown with and without the slope modification with a slope modification defined as a positive slope observed on the 6-month model.
| Model w/ Slope | Model w/o Slope | ||||
| -1 | -2 | -1 | -2 | ||
| Avg | 14.44% | 66.78% | 13.00% | 42.14% | |
| Median | 9.64% | 66.78% | 9.64% | 40.39% | |
| Max | 79.91% | 79.91% | 88.09% | 88.09% | |
| Min | -8.42% | 53.65% | -46.95% | 2.77% | |
It appears that adding a slope modification vastly reduces the risk of a significant loss in the historical experience. This is as the minimum downside is vastly reduced with the slope modification.
I am continuing to test various time periods and base model used in the slope calculation but this risk control seems promising. Presently, the 6-month slope is positive and in conjunction with the models suggest that positions in gold/precious metal shares should be accumulated. How does this differ from my call a couple months ago. For starters, the 6-month slope was negative then.
No comments:
Post a Comment