First lets review the formula.
A correlation between two variables is the covariance between each divided by the product of each variables standard deviation (or the square root of each variables variance).
Since the correlation is a normalized number it is important to remember that it is a user friendly but not that usable variable. It describes the co-movement but says little more than that. There are much more powerful analytical co-dependence functions than correlation. Product advertisement: we are thinking about developing a powerful yet fun to use platform that will facilitate statistical modelling of financial markets.
Anyways, look at the recent rolling correlation of the S&P 500 and the US dollar. This is a rolling 1 year correlation using weekly percent return figures on SPX and DXY.
- Notice the very positive correlation leading up to the Tech bubble market peak in early 2000, where the correlation was as high as 60%. Here the US economy was booming, the stock market was in a dizzying rally, and the US$ continued to strengthen (the EUR was at $0.85 in 2000, versus $1.35 now).
- The average correlation in this period was -20% as a strong dollar usually meant bad news for the US economy as it hurt exports, and correspondingly the market. Leading up to August 2008, the equity market was correcting and the equity market was selling off until September when Lehman went broke, and there was a massive flight to quality and the US$ rallied. This threw the market’s negative correlation back to nearly all time low of -60% until …
- From less correlated to slightly less correlated. The correlation is increasing. We had a substantial bottom in the Dollar in December 2009 (since then the DXY is up nearly 8% and SPX is up about 6%).
It is important to mention that though the trade weighted Dollar is up 8% since early December, it is up almost 11% versus the EUR. This implies that versus other currencies the USD has not gained much.
Thanks for reading,
The Covert Analytics Team














