8/26/08

Does What People Search For Have Predictive Value?

Vlada at StockWeb has posted an interesting article on which keywords attract blog visitors from search engines at different times, according to how the market is doing. I've noticed the same trend at my own blog.

It seems to me that visitor trends may have some predictive value. It's sort of conventional wisdom that retail investors try to get in on a trend just as that trend has run its course. (And in hindsight, symptoms of trend reversals are easy to spot. For example, a number of companies funded share buybacks and dividends by borrowing money in 2006/2007. That was a great time to short the market.)

For instance, visits to this site from readers looking for bear market ETFs peaked around March 8th, 2008. The market raced up over 10% from that point until mid May. The same thing happened in late January through early February, when the market reversed course and went up.

Visits from people looking for information on investing in agriculture seem to peak when uptrending commodity prices are about to reverse course.

Of course, search engine ranking, which changes constantly, has a lot to do with how many people come here looking for a particular topic. For example the number of people searching for inverse ETFs may be constant, but as my site moves up and down in the search results, the number of visitors waxes and wanes.

But I think there is something to the predictive value. In the summer of 2007 I was still contemplating legal work. Over two thirds of the jobs advertised were for mergers and acquisitions. That's when M&A activity started drying up.

So, perhaps it'll be a good idea to start investing in stocks when the majority of job ads on legal career sites are for bankruptcy lawyers? It would be interesting to see.

While this site's number of visitors has gone up dramatically since it was first started in December 2007, the number is still fairly low. So my visitor peaks and troughs for various topics may not be very accurate in predicting market trends, if they have any predictive value at all.

If I have time (school's starting soon), I'll be using Google's Adwords tool, which provides approximate monthly search volumes for keywords of one's choice. Just offhand, there were approximately 1,300 searches for inverse eft in July, significantly below the average volume of 2,900. July 15 marked this year's stock market low (so far). Too bad Google doesn't provide daily figures. I'd be curious to see what portion of the search volume came before the 15th, and what portion came after. We'll see what August's figures have to say.

A few things about Google's search volume for particular keywords. Google isn't the only search engine out there, and its trends may be different from other search engines. I don't think this will make the results inaccurate though.

There's also the concern that those who search for inverse ETFs aren't necessarily bearish--maybe they're neutral, or even bullish (maybe they want to short the ETFs or buy puts). I don't think this one will skew results much.

What can skew results, however, is that monthly searches for things like inverse ETFs most likely leave out all those bearish investors who have previously searched for the term. For why would they search for the same thing twice or more when they likely obtained all the information they needed the first time?

It's most likely a silly fantasy, but it would be cool if search engine volume could be used to make a sort of optimism/pessimism index similar to one the main character in Galbraith's satire used to make a lot of money.