Correlation of S&P 500 to Oil Prices

It is a long standing belief that when oil prices are up the market is down, and vise versa.

Below we highlight the correlation of changes in the S&P 500 to changes in the price of oil, with oil prices plotted as well. Over the long term, since 1988, the correlation coefficient was positive 44% of the time, negative 56%, and averaged -0.05.

click to enlarge

Comments

[oiltrader:]

click to enlarge?? click WHAT to enlarge?

2007 Dec 06 11:09 AM

[Giovanni di...:]

These statistics show whatever you want to see. negative correlation, positive correlation. We discussed this articles among some critical investors and maybe you want to answer to this comment:
----------------------...
Interesting topic but I don't know what they measured.

Was correlation measured instantaneously? at the end of each trading day (oil up a dollar S&P down a dollar, etc.)? monthly? quarterly?

Even if there was no simultaneous variation over whatever interval they were measuring it is entirely possible that there was correlation with a lag period, e.g., S&P might go down 3 months after a sustained rise in oil price, etc.

The theory behind the graph they showed is terrible; overlaying an independent variable on the dependent variable adds no information at all, yet the eye tends to view the rising oil prices of late as somehow reinforcing the absence of consistent S&P correlation.

Drug companies and less reputable researchers do this sort of thing all the time, designing visuals so as to suggest whatever outcome they deem favorable instead of showing the original data, a practice I used to rail against in my academic days. Why not just superimpose S&P against time and oil price against time? After that basic data is displayed one can get fancier with statistical graphs and the reader will be able to judge for him/herself whether the statistics seem to fit the original data or are a stretch.

2007 Dec 07 04:15 PM

[Emanuele:]

Hi Giovanni,

My name is Emanuele and I am doing a Master in Corporate Fiancne and Investment. I am contacting you after carefully reading your comment - which I compeltely agree to by the way - to ask you how exactly you would go about measuring correlation between oil prices and stock prices over time. To cut a long story short, I am about to investigate the correlation between crude oil price and stock price of twenty clean energy companies from the Wilder Hill Index ... I'd really appreciate if you could give me a piece of advice on this... rigamonti.emanuele@goo... is my e-mail address.

Thanks in advance Giovanni (Shall I say grazie mille...???).

I look forward to hearing from you,


Emanuele


On Dec 07 04:15 PM Giovanni di Maggio wrote:

> These statistics show whatever you want to see. negative correlation,
> positive correlation. We discussed this articles among some critical
> investors and maybe you want to answer to this comment:
> ----------------------...
> Interesting topic but I don't know what they measured.
>
> Was correlation measured instantaneously? at the end of each trading
> day (oil up a dollar S&P down a dollar, etc.)? monthly? quarterly?
>
>
> Even if there was no simultaneous variation over whatever interval
> they were measuring it is entirely possible that there was correlation
> with a lag period, e.g., S&P might go down 3 months after a sustained
> rise in oil price, etc.
>
> The theory behind the graph they showed is terrible; overlaying an
> independent variable on the dependent variable adds no information
> at all, yet the eye tends to view the rising oil prices of late as
> somehow reinforcing the absence of consistent S&P correlation.
>
>
> Drug companies and less reputable researchers do this sort of thing
> all the time, designing visuals so as to suggest whatever outcome
> they deem favorable instead of showing the original data, a practice
> I used to rail against in my academic days. Why not just superimpose
> S&P against time and oil price against time? After that basic
> data is displayed one can get fancier with statistical graphs and
> the reader will be able to judge for him/herself whether the statistics
> seem to fit the original data or are a stretch.

Apr 02 12:55 PM