How do I find investments that are negatively correlated?

April 12, 2004 12:27 AM

How do I find investments that are negatively correlated? [lots more inside]

Now that I've got some money for the first time since I took Economic Statistics, I'm trying to recall what I learned, at least for fun and possibly profit. In particular, I recall learning that you could take two negatively correlated investments – one more volatile but with higher expected return, the other less volatile but with lower expected return – and come up with a mixture that would give you a higher return and lower volatilaty than the lower volatilty investment alone.

So I'm wondering where one looks in the real world for these kinds of investments. I've watched with interest as mortgage-related investments and to some extent other real estate have done extremely well while stocks and staid banking investments haven't, and have been tempted to conclude there's something fundamental at work here, but still feel like any understanding I have on how different economic sectors interlock is pretty naive. Are there sectors whose fortunes are generally reversed? What are resources I can use to learn more about this?

Also, I suppose in general if I'm going to get serious about this, or even get serious about dabbling/studying/understanding this I'm probably going to need some good dataset and crunching tools. Any suggestions?

posted by weston to work & money (6 comments total)

This answer may not be in the spirit of askmefi but I was always under the impression that the technique you mentioned was more academic than practical. If it was easy everyone would do it.

That said, this is my recommendation:
Look at multiple investment vehicles.

You may have difficulty for example, trying to find two stocks which are closely negatively correlated. If you look at multiple types of investments though the task might be easier.

Stocks vs. Bonds is the classic scenario of course.

You may also want to look at futures. I bet you could get a pretty decent negative correlation between returns on the oil futures market and airline stock performance.

Good luck, I'm interested to see what suggestions come up in this thread.
posted by yangwar at 6:51 AM on April 12, 2004

You gotta pay to play, weston.

Buy your data from Bloomberg, and a full featured MVO from whomever is selling them nowadays.

You could always buy floating rate bonds and inverse floaters, which are perfectly negatively correllated, but then you'd just have a fixed-rate bond, y'know?
posted by Kwantsar at 9:23 AM on April 12, 2004

Also remember that assets which show little or negative correlation in uncalamitous markets often find correlation during crashes and sharp downturns.
posted by Kwantsar at 9:43 AM on April 12, 2004

I think the classic example of this sort of thing is diversifying a stock portfolio to include foreign stocks. Over the long term, a diversified portfolio that includes an EAFE fund (foreign index fund) and a S&P 500 fund gives better returns and has less risk than one of those funds on its own.

I don't know that anyone has ever broken this down to a more micro level. No one, to my knowledge, has conclusively proven that, say, when gold mine stocks go down, underwear stocks go up. But if you deal in much larger categories--"U.S. Stocks" "S&P 500 Stocks"--correlations are easier to spot.

While it is possible to roll your own diversified portfolio, beginning investors are usually best off buying a balanced mutual fund. They are imperfect instruments, but good for people who don't want to bother.
posted by profwhat at 1:01 PM on April 12, 2004

I don't know that anyone has ever broken this down to a more micro level.

They have, but that research takes work, and has costs. You see an asset class or sector correlation or covariance matrix in Barron's or WSJ every now and then, but what Weston's looking for has to be DIY, unless he wants to pay Ibbotson or his local pension consultant to do it for him.

Also, beware false prophets. If you can, take out Bogle's "A Random Walk Down Wall Street" from your local library, and read the story of Mead Johnson, which moved inversely to the market purely because of chance, and Metrecal.

Kwantsar is probably right. Call Bloomberg and I bet they can ship you a CD with many historical prices. You'll have to provide the computing yourself to find the correlations, but the formulas should be available in a college-level stats textbook.
posted by trharlan at 1:21 PM on April 12, 2004

There are actually mutual funds that are negatively correlated. The Inverse Profunds are basically public hedge funds that are supposed to decline when the market increases, and vice versa. Of course, past performance is no guarantee of future performance. Be warned that the funds do encourage active trading (otherwise, you could lose all your money) and as such, there management fees are in the 2-3% range, which is much higher than the Vanguards of the world at 0.40% or less.
posted by calwatch at 10:08 PM on April 12, 2004