Pages

14 January 2012

Implosion of the inverse correlation in the markets?
To measure is to know

A correlation comparison study on USD/EUR, SPX, Gold, Oil, AUD/USD and VIX

The other day a visitor on my blog stated the markets inverse correlation with the US Dollar had pretty much imploded. In the meantime AlbertaRocks also published his peace on this subject: USD - "Inverse" About To Become The Norm
To be able to validate this statement in a better way than eyeballing the charts, I wrote a thinkscript study that compares the trend in a given stock, index or currency against another showing the correlation, be it positive or negative, in a subpane. That is: when both rise or decline in tandem, positive correlation is at hand, signaled by a green flag. Otherwise negative correlation is assumed, signaled by a red flag. Because of data issues I had to use the inverted EUR/USD spot prices as a proxy for the US dollar.

SPX : USD/EUR
My measurements show that on a week to week basis green and red flags fly all over the place in an erratic, chaotic way. On a day to day basis the randomness seems to be even greater. So any meaningful correlation on these timeframes is virtually non existent, neither negative nor positive (see upper subpane on charts).

When correlation is measured by the waning and waxing of the weekly 10-period moving averages (50-period daily moving averages), the most extensive conclusion in my judgement would probably be that there are times when values move in tandem or non-tandem for a while, but both connections and disconnections seem to "just" happen. Whether the markets rise or fall, is of no difference too. And tandems fall apart just as quick as they pair up together, seemingly for no obvious reason.

Only when measured on the larger timeframe of a 40-period moving average (200 days) the quite popular believe that there was inverse correlation until recently turns out to be only partly true. My TOS-charts go back 20 years and until April 2002 positive correlation had the upper hand. Since then inverse correlation took over as the prevailing trend. But notice also during these recent years there are weeks and weeks on stretch with positive correlation.


Gold : USD/EUR
This "pair" shows even on a week to week timeframe signs of inverse correlation. Here the "norm" of inverse correlation seems to have a sound basis, but extended periods of connected price movement are no exception either.

Oil : USD/EUR
The negative correlation is less distinct as in Gold. On longer timeframes disconnection shows stronger.


SPX : AUD/USD
Since 2002 there is quite a strong bias for prices to move in tandem, but here too the aforementioned exceptions apply.

SPX : VIX (inverted)
In this "pair" we find the fairly consistent correlation most traders are looking for. The dominant positive correlation is proof of the axiom that complacency rules at market tops, while fear is dominant on market bottoms. It shows contrarian thinking pays of for smart money at the expensive of "dumb" money. Think or Swim!


So did the inverse correlation between the dollar and the markets "implode"?
Sure it imploded, it implodes and explodes all the time, but on the longer time frame non-tandem behavior still seems to be dominant.

My conclusion tends to be for shorter time frames that correlation resembles more like noise on an empty radio channel than for an explanation for market developments. The statement "SPX was higher today, because the dollar dropped" is as often true as it is untrue. So it has no value in itself. In my opinion it has more to do with chance, like flipping a coin, than with market analysis. I postulate it of no importance whatsoever for day traders or intermediate term swing traders. For long term implications I refer the interested reader to AR's excellent post on his blog (See blogroll).

[The thinkscript code of "TrendCorrelation" is posted in the comment box below]