The gist of this trade is that I believe the variance between gold and silver is just background noise in what is, at its core, a coupling that have a stationary relationship.

The point of the post is that I believe the easy money in riding the momentum of gold and silver is behind us. I do not think many people on the planet have an information edge on the direction of precious metals anymore.

The trade is (and the calculator will tell you when a trade should be made): If the relationship between gold and silver strays 25% in either direction from what I have calculated to be the best "mean" relationship, you go long and short in such a manner that you make money if the relationship normalizes, at which point you close both positions.

So for instance, if gold rallies substantially, but silver is stagnant, we would buy silver and short gold.

The benefits of this are two-fold. First, as long as they normalize you will make money, it does not matter if gold goes back down, silver goes up, or they meet in the middle. Second, as has been pointed out, this trade is hard to predict. Because you are long one and short the other, you do not need capital on the sideline for this trade. You use the short proceeds to go long.

As was pointed out by previous commentors, this trade requires patience and diligence. That is where the alpha in this trade is.

If it is still not clear I recommend making a copy of the calculator for your own use and playing with the values "Price of GLD" and "Price of SLV". The calculator will let you know when a trade is on. The only thing you have to remember to do is close the positions when the gap returns to +/- 5% of the expected. The paper I cited earlier in the comments waited for a return to normalcy and had good results, but I prefer to be conservative.]]>

The gist of this trade is that I believe the variance between gold and silver is just background noise in what is, at its core, a coupling that have a stationary relationship.

The point of the post is that I believe the easy money in riding the momentum of gold and silver is behind us. I do not think many people on the planet have an information edge on the direction of precious metals anymore.

The trade is (and the calculator will tell you when a trade should be made): If the relationship between gold and silver strays 25% in either direction from what I have calculated to be the best "mean" relationship, you go long and short in such a manner that you make money if the relationship normalizes, at which point you close both positions.

So for instance, if gold rallies substantially, but silver is stagnant, we would buy silver and short gold.

The benefits of this are two-fold. First, as long as they normalize you will make money, it does not matter if gold goes back down, silver goes up, or they meet in the middle. Second, as has been pointed out, this trade is hard to predict. Because you are long one and short the other, you do not need capital on the sideline for this trade. You use the short proceeds to go long.

As was pointed out by previous commentors, this trade requires patience and diligence. That is where the alpha in this trade is.

If it is still not clear I recommend making a copy of the calculator for your own use and playing with the values "Price of GLD" and "Price of SLV". The calculator will let you know when a trade is on. The only thing you have to remember to do is close the positions when the gap returns to +/- 5% of the expected. The paper I cited earlier in the comments waited for a return to normalcy and had good results, but I prefer to be conservative.]]>

A mean reverting strategy does not require a measure of correlation or causation, only that the mean reverts. Volatility is what allows this trade to be profitable. If there were not relatively high volatility there would be no trade to make.

@cozgenc

Using an Augmented Dickey-Fuller (using the difference of the log values of the price time series) test we get a p-value of 0.01 and a test statistic of -5.69 with a lag order of 7. In other words it passes the Dickey-Fuller test for cointegration extremely well, and we can infer stationarity of the price series.

The equation used comes from the paper, "Pairs Trading, Convergence Trading, Cointegration", by Daniel Herlemont.]]>

A mean reverting strategy does not require a measure of correlation or causation, only that the mean reverts. Volatility is what allows this trade to be profitable. If there were not relatively high volatility there would be no trade to make.

@cozgenc

Using an Augmented Dickey-Fuller (using the difference of the log values of the price time series) test we get a p-value of 0.01 and a test statistic of -5.69 with a lag order of 7. In other words it passes the Dickey-Fuller test for cointegration extremely well, and we can infer stationarity of the price series.

The equation used comes from the paper, "Pairs Trading, Convergence Trading, Cointegration", by Daniel Herlemont.]]>

My reasoning is based largely on the writing of market monetarist Scott Sumner, who runs an excellent blog at http://bit.ly/HEbYti .

There is of course a real cost, it is opportunity cost.]]>

My reasoning is based largely on the writing of market monetarist Scott Sumner, who runs an excellent blog at http://bit.ly/HEbYti .

There is of course a real cost, it is opportunity cost.]]>

The optimism-pessimism spectrum was only meant to be a tool to help visualize where a trade opportunity might lie. The point I was trying to make was that it is not necessary to foresee a shift from optimistic to pessimistic. A shift from very optimistic to optimistic would work as well.

This is because I believe the unexplained residuals in the first graph can be better explained when considering expectations of the market (whether wrong or right), and that when there is a shift in expectations there is also a shift in which line best represents the correlation (of the two in graph 2). This is as opposed to a simple shift in the price of the dollar, where we simply move along whichever line we happen to be on (determined by sentiment).]]>

The optimism-pessimism spectrum was only meant to be a tool to help visualize where a trade opportunity might lie. The point I was trying to make was that it is not necessary to foresee a shift from optimistic to pessimistic. A shift from very optimistic to optimistic would work as well.

This is because I believe the unexplained residuals in the first graph can be better explained when considering expectations of the market (whether wrong or right), and that when there is a shift in expectations there is also a shift in which line best represents the correlation (of the two in graph 2). This is as opposed to a simple shift in the price of the dollar, where we simply move along whichever line we happen to be on (determined by sentiment).]]>