There are many articles about the dependency of the gold price on the U.S. bond yield, the Dollar Index, and the overall situation on the stock market. But I personally have not seen a specific statistical model that attempts to predict the price of gold on the basis of the simultaneous impact of the above mentioned factors. I tried to fill this gap and got interesting results worthy of your attention.
I set myself a goal: to build a linear multifactor static model that predicts the gold price on the basis of the three independent variables:
- The U.S. 10 Year Bond Yield
- The Dollar Index
- S&P 500
I used daily stock data from January 2015.
Let me start with the cross-correlation matrix, demonstrating to what extent the selected independent variables are correlated between themselves and the price of gold:
As we can see, for the specified period of research, the greatest impact on the price of gold had: the U.S. 10-Year Bond Yield (corr = -0.81) and the Dollar Index (corr = -0.45). The impact of the stock market (S&P 500) was less significant, however, it still took place. And, by the way, in this case, the dependency was direct (corr = +0.34).
It is important to note that selected independent variables are not strongly correlated between themselves, which is an important prerequisite for any qualitative regression model.
Using the R programming language, I got the formula of the desired model with the following coefficients:
Note that the model determination coefficient (R2) is above 82%, which indicates its high forecasting quality.
And now the most interesting part. Below you may find the histogram of deviations of the actual gold price from the predicted, obtained with the formula:
As you can see, despite rather strong decrease in the gold price observed since August 2016, the model indicates that the current gold price is still above its sustainable level by more than one standard deviation.
For a better understanding of the obtained results, please take a look at the following chart which shows the actual and the predicted prices of gold:
And here is the deviation chart of actual prices from predicted prices:
It should be noted that, firstly, whenever the distance between the actual and the predicted price was more than one standard deviation for a long period of time, it was followed by a local reversal in the dynamics of gold.
Secondly, since October, the current price of gold has been above the price, predicted by the model. It indicates that the gold market is consistently overbought against the background of the current dynamics of the U.S. bond yield, the Dollar Index and the U.S. stock market.
In the beginning of the post, I briefly explained that the dynamics of the U.S. bond yield, as well as the Dollar Index, have the greatest impact on the gold price at the moment. Given the expected tightening of the U.S. monetary policy in 2017 against the background of zero and negative rates in Europe and Japan, I do not see any prerequisites for the prompt decline of the dollar value. On the other hand, against the backdrop of an uncertain Trump policy, in my opinion, there is no reason to expect the decline in the U.S. bond yield in the foreseeable future.
Based on my assumptions and the obtained model, it must be concluded that gold is deprived of growth drivers and is still above balanced level, which indicates a high probability of further reduction in its price.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.