High grade copper has been on a reasonably flat trend in recent years, with the current price of $2.72 being 40% below the peak of $4.62 and 45% above the low of $1.91 over the last six years. This year, it has been among the under-performers in the major commodity cohort-
The volatility of the copper price has tended to move in line with that of the DJ UBS Commodity Index, lying about 50% above it -
Given the wide variety of its applications, it is often suggested that the trends in the copper market are a useful leading indicator of the state of the world economy. Over the last six years, the correlation coefficient between the copper price and ACWI, the All Countries World Index six months later has been 0.25. This suggests that changes in the copper price are a reasonably strong world economic lead indicator.
We have developed a fair value of the copper price based on a multiple regression analysis of its price on 22 driver variables including interest rates, stock indices, other commodities and exchange rates. The database is six years of daily prices.
The elasticity of the copper fair value to groups of driver variables is shown in the graph below.
As can be seen, copper is negatively correlated to the dollar, as is the commodity index.
We have also developed a prediction model based on the lead given by the change in various indices we believe to be related to the variables in the regression model referred to above.
Based on the lead given by these indices, it is possible to develop price predictions for a range of commodities, interest rates, stock indices and currencies. These can be combined with the fair value model elasticities shown above to give a predicted movement in the copper price over the next month. We also back test the prediction models to assess their value in a trading environment (taking a long position when the predicted price change is positive and a short position when negative).
The predicted price changes for the next month, and the annualized return/volatility ratio over the last six years from trading the predictions, are shown in the following table –
The 127% return/volatility from trading all the predictions combined compares with that applying to the S&P 500 over the same period, of 101%. We consider this to be reasonably robust given the period commences in mid-2009, around the start of the latest bull market in US stocks.
Our predicted movement in the copper fair value for the next month is shown in the table below. The contribution of each driver variable to the overall decrease is also shown.
As can be seen from the above two tables, our positive view on the US dollar over the next month plays heavily into the predicted copper price movement.
We also used the index model to directly forecast the copper price movement for the next month. A change of –1.6% resulted, confirming the signal given above. The back test return/volatility ratio from trading this model was 54.1%.
Our fair value estimate of the copper price (using the 22-variable regression model) is 8% below the current price.
Despite the volatility of the copper price, allocating a component of a US stock portfolio tracking the S&P 500 to copper traded using the index model referred to in this article would have improved the return/volatility ratio of the portfolio. For a 5% allocation, the ratio would have improved from 101.2% to 104.2%.