I have to admit this one went right under my radar until recently. FT.com put up an interesting and amusing (as you will see) article on matching stock price against Google Correlate.

I leave it to the reader to look over the Google Correlate FAQ but in short it finds the closest match for a time series (dates against values) and frequency of search activity. In other words we can put in stock prices for a period of time and it will find the “closest fit” for search terms. As Google warn “Correlation is not causation” but it can be illuminating to try.

Anyway, I love this kind of thing so I thought I would plug a few time-series for some of our favourite technology giants and see what came up. I haven’t put a link to the time-series files as I am unsure what Yahoos policy is about publishing this data but I’ve put instruction on how to get it and format it. If you are clueless email me and I will send the files.

Apple Inc

To start off lets get the Apple stock prices data from Yahoo finance. Go to Yahoo Finance, search for AAPL and then select “Historical Prices” from the menu on the left. This will take you here. Download the CSV file spreadsheet.

If we open the speadsheet we have a number of columns.

The format that Google Correlate takes is Date|Value so we have to choose a column to plot. I am no financial whizz but I reckon closing price “Adj Close” will do so we’ll edit the file to just have these two columns.

Next we have to alter the date column to be in the format which GC takes so reformat the date column to “YYYY-MM-DD”. Don’t forget to remove the header and save it in CSV format.

Lets plot!

The found correlations are:

I’m going to choose the top correlation in each example. In this case as you can see it is “smartphones”.

I am going to go out on a limb here and say that in this case there is a link from the search results to the share price. From the iPhone launch in 2007 we can see the stock price make a sharp rise as the “early adopters” started to query what exactly a smartphone was and liked what they saw.  In 2009 (according to Wikipedia) the “there is an app for that” campaign started so that might account for the greater interest from the general public not savvy yet to the new “big thing” and getting out their credit cards to push Apples fortunes up up and away.

Google

The found correlations are:

Uh, right. Lets plot!

St Louise Backpage is a classifieds site. One to watch clearly for Google shareholders. Maybe time to start putting all that junk you were saving for the charity shop up for sale instead.

Microsoft

The found correlations are:

Plot.

If “Google Interview Questions” means what I think it does then perhaps Microsofts steady (relatively speaking) share price might indicate a steady outflux of engineers looking to find out the secrets of gaining entry to Googles superior canteen. And, why the massive spike in 2010? Androids sudden take-off perhaps.

Hewlett Packard

To be fair I am not sure if I have focussed on the correct HP division. This is the company stock.

The found correlations are:

Plot.

Frankly I am not sure what Ekas Portal is but it’s maybe time a few HP shareholders booted up MS-Paint and joined in as there is it may just boost their share price. Something to do with HP printer prices perhaps??

Conclusion

Finding causality between the fortunes of a company and Googles search terms does sound good on paper. In reality it is clearly a bit of a minefield. It’s an intriguing idea. I would love to see if anyone can come up with some good correlations that make sense.

One final note. It’s interesting to see that all the companies had a dip in 2009. Why?

Thanks to  quantly for pointing me to this article and Curated Alpha for finding it in the first place.