Commodity Correlation Table

Discussion in 'Ag Futures' started by runningman, Feb 21, 2007.

  1. Has anyone ever seen a table with all or most of the commodities on it with the corresponding correlations with each other?
     
  2. Wow! Corn/Dow! 91% correlated!
     
  3. That is very interesting..

    I'm going to study that chart very closely..
     
  4. The Dow Jones and Corn have both trended higher during the past 6 months. It's a random coincidence that they have a 91% "correlation". Corn and Soybeans having a 93% correlation is a real correlation.
     
  5. Nick Leeson Jr: That site has been bookmarked. Excellent find! Thank you!

    Regarding the linked graph, make sure you guys understand what you are looking at. It's interesting, but it's not historical...little more than hindsight. They are a for-profit subscription service!

    As mentioned, I'd agree the Corn/Dow correlation is nothing more than random. Likely other random correlations in there too.
     
  6. You people don't know what you're talking about.

    Dow component companies employ people.
    Dow going up results in people being paid more.
    Well-paid people can now afford to substitute high-grade Corn for the cheaper feed they usually eat.
    Therefore Corn goes up.

    Economics 101.

    Frankly, I'm surprised the Dow/Coffee correlation isn't higher. Just think of all that free cash flow that Starbucks should be on the receiving end of!
     
  7. KS96

    KS96

    There are no such a thing as a "random correlation".
    There is either correlation or not.
     
  8. Not true at all. It is perfectly appropriate to apply parametric or non-parametric tests of statistical significance to a correlation coefficient.

    The link above refers to a large, 35x35 correlation matrix, with 35x34=1,190 elements. By pure chance, at any given moment in time, we can expect at least dozens of high numbers relatively near 1, as well as dozens of low numbers relatively near -1. That does not necessarily indicate any statistical significance of the correlation coefficients for any of those pairs.

    If we were to go either back or forward to another moment in time (sufficiently far away compared to the length of the correlation period itself, here 180 days), many of those high absolute value correlations will disappear, while many others will arise. That's the essence of randomness.
     
  9. KS96

    KS96

    Says who?
    Maybe you can apply a statistical significance test
    to your statistical significance test. Where does this stop?
    Everything allowed in the books, but my
    experience says that you shouldn't ask whether a
    correlation exists by chance or not.
     
    #10     Feb 26, 2007