Bitcoin Forum

This thread is intended as a refutation to any and all posts that use Google search volume as an indicator of investor sentiment in order to predict future price trends. Google search volume is NOT a reliable indicator of future price trends, as the following chart will show.

Note: I originally intended to show that the price did reflect prior search interest and was delayed by a few days due to bank transfer times, but what I discovered turned out to be the opposite.

First I had Google Trends show the search volume graph for "bitcoin" selecting "last 90 days" for the time scale. If you mouse over the graph at various points, a little infobox will pop up showing you the date for that point:

http://anonmgur.com/up/3568e76b33cec63346522d5fcdb30178.jpg

The very first date on the far left of the graph is January 22, 2013. Then I moused over the end date:

http://anonmgur.com/up/e300c8fcc299601c9f6f1f0e6ff405f8.jpg

The last date shown on the graph is April 19, 2013.

So I went over to Bitcoin Charts (http://bitcoincharts.com/) and selected a custom date range with those same start and end dates:

http://anonmgur.com/up/d05b362f1c28d04b46635180b3cd7443.jpg

And generated the Mt. Gox price chart. Since the price chart and the Google chart were different widths, I had to measure the exact with of the trend line in the price chart (which was wider) which was 855 pixels. Using an image editing program I selected only the search volume trend line from the Google chart and scaled it to exactly 855 pixels, so the length from start to end was the same.

Below is the superimposed Google search volume chart over the Mt. Gox price chart:

http://anonmgur.com/up/1219c566844a8b244271c4794b6f576d.jpg

The search volume follows the price by around 24-48 hours. The epic crash occurred the morning of Wednesday, April 10. But search volume continued to rise throughout Wednesday and peaked on Thursday, April 11 (you can verify this on Google Trends yourself). In fact, on Tuesday, April 9, when the BTC price was at its peak, the search volume was only 57% of what it was 2 days later, well after the crash. This seems to suggest that the media typically reports stories in response to increasing price, which spurs public interest, but that public interest doesn't result in a spur in demand.

Additionally, the smaller spikes in the search graph seem to trail the spikes in the price graph by around a day. Spikes in the search graph, however, are not typically followed by spikes in the price. Towards the center of the graph, there is a small spike in the search volume, but there is no subsequent spike in the price. Around the April 06 mark, there is a swift drop in the search volume, without any concurrent or subsequent drop in price. This indicates that search volume is not a reliable predictor of price.

Interesting, some real cause-and-effect in the speculation forum.
I wonder why the google sreach volume is sinking so much in the last part though, more people know about Bitcoin than before.

Maybe that's precisely the reason. Now that everyone knows about it, they don't need to search it.

Thanks for the great comparison chart. I regard Google search trends as a following, and confirming indicator for bitcoin price movements.

The comparison is very telling, however here is an important question: is either graph smoothed using a moving average? If so, then the result of your comparison is invalid (unless they both use the same moving average).

For example, if the google trends graph uses a 5-day moving average, then its graph will be shifted to the right by 2-1/2 days, producing the result that you see.