Online:  
Visits:  
Stories: 
Story Views  
Now:  
Last Hour:  
Last 24 Hours:  
Total: 
Matching the HUI miners index to the gold price graphLet's start off with a graph scaled to match the Gold price (left axis) to the HUI miners index (right axis).
Gold (red, left scale in USD/Oz) and HUI (blue, right scale); Data till Nov 16, 2016  click to enlarge 
The time scale extends from July 2013, as gold started recovering after its major slide since April 2013 till the present day. 28 points on the HUI index matches a $50 variation on the gold price. One very important observation though: the 'zero level' on the HUI would match to a reading of $900/Oz for gold.
This fresh rule then is written as: HUI = 0.56 * (Gold price – $900)
This essentially makes the relationship nonproportional and implicitly assumes that a gold price below $900/oz would imply the gold miners to go out of business. A disclaimer not without any reason: there is no such clear 'cutoff' gold price for the mining industry. The level of the USDindex highly affects mining costs for the majority of gold producers outside the US. Moreover, gold producers have vastly different margins and most probably also high margin differences among their individual gold mining sites. Yet a rule with only two parameters is most elegant … while the current correlation holds.
Least squares regressionParameters in the above equation were rounded to make major divisions on the left (Gold) and right (HUI) axis coincide. Both the gold slide during the latter part of the bear market as the subsequent 2016 recovery are well matched by the concurrent plunge followed by the magnificent rally of the miners. The same accounts for the correction we have witnessed since mid August 2016, whereby recovery gains eventually may be at stake.
A (least square) regression analysis provides you with mathematically exact values. For such least squares linear regression, we opt to extend the interval of the gold slide from about $1800 to the $1050 bottom early December 2015. To cover this interval, we need including data going back to the summer of 2012, before gold failed its last attempt to break above $1800 early September 2012. Any time interval chosen necessarily yields a slightly different set of parameters, however including these earlier data even improves the correlation coefficient to 0.968. This corroborates the analysis results.
July 2012 to Nov 2016: Gold – HUI spread (blue dots) and regression trend line (red). (click to enlarge)
'Early' September 2011 to June 2012 (Gold, HUI) values are shown in orange.

In a regression analysis, dates are eliminated: it shows the HUI values relative to the gold price. All blue dots are given (HUI, Gold) couples for any given date between July 1, 2012 and Nov 16, 2016.
Note: The regression made is yvalues: Gold price on xvalues : Hui index. It provides the intercept on the gold price axis and the inverse value of the slope of HUI relative to gold.
Using these data, the slope and intercept are 0.555 and $893, turning the HUI dependency on gold into
Why we don't include any data going back to the September 2011 gold price all time high also proves obvious. Those earlier data couples are shown in orange. When checking carefully (clicking the graph for full resolution), those data points are in a cloud at the upper right, systematically above the points since July 2012. Including those data tilts the regression line higher, but it deteriorates the correlation coefficient. During 2011 – when the HUI failed to catch up with the gold price steaming up to its August/September all time high – the HUI/Gold relationship was undergoing its major paradigm shift. The 'bearmarket logic' only set in after gold made a few more failed attempts to break and hold above $1800.
Analysis of the residualsWhile the trend line gives an overall decent fit, data points in the intermediate area are scarce and seem to concentrate on the downside, while the data cloud at the right is quite extended. For a meaningful explanation, we need to introduce the dates once again, showing the 'residuals' as a function of time. These distances from the least squares trend graph are shown as a function of time on the below graph:
Residuals as a function of time from May 2012 till Nov 16, 2016 (click to enlarge) 
Residuals are oscillating around the trend line. In 201213 the amplitude of the oscillations used to be quite large, giving rise to the more extended data cloud preponderant at higher gold prices. After the gold price slide, their amplitude is reducing in 20142015. The data cloud now is more narrow. Below you find the more recent data.
Residuals as a function of time from July 1, 2015 till Nov 16, 2016 (click to enlarge) 
In October 2015 miners ignored the temporary gold recovery and residuals went negative. Yet they reverted course and became positive before gold eventually bottomed at $1050/Oz early December 2015. With gold dragging above its bottom throughout December, residuals slid and became negative into 2016. As gold eventually started its recovery after mid January, residuals slid even deeper, indicating that gold miners reluctantly followed gold. The lack of enthusiasm among mining investors was comparable to that during the fake October 2015 recovery. Sentiment would only start improving after mid February and it lasted till early April 2016 before any positive reading on the residuals confirmed that the recovery trend was picking up steam, finding acceptance among a larger number of investors.
Residuals remained positive from mid May 2016 (despite a late June trough) throughout August 23. By mid August gold had started weakening, with the slide accelerating towards the end of the month. This eventually killed miner enthusiasm. Residuals remained negative after Aug 30 till Oct 12. Recently, they have been oscillating between +10 and 10 (more or less market neutral).
A negative reading implies that mining investors on average accept selling their mining shares at lower prices than what the 'regression rule' suggests. At any positive reading, investors are ready to pay more. The sign and trend of the residuals is not meaningless at all. The below description is 'a caveat' against considering negative residuals a strong 'buy signal' for miners.
A persistent negative reading sometimes is the prelude for a slide of the gold price. A 'textbook example' is shown in the graph below, where the residuals are plotted together with the gold price before and after the hectic gold selloff in April 2013 continuing to June 2013.
Residuals (red dots, left scale) as a function of time from Christmas eve 2012 till July 26, 2013, The gold price in USD (blue graph) is shown on the right scale. (click to enlarge) 
Earlier articles on this topic