To Buy or not to Buy? How Google Trends Data Relates to Bitcoin Price

Hazel Monks
HODL Finance
Published in
4 min readMay 2, 2019

Key takeaways:

  • The relationship between Google trends data and the price of bitcoin is one which has been discussed as early as 2011
  • A recent study has demonstrated that price increases in bitcoin drive increases in “Buy Bitcoin” google searches as opposed to the other way around

Several news outlets have recently been reporting on the relationship between the price of bitcoin and Google Trends data for users searching “Buy Bitcoin”. The statistical tie between Google Trends and Bitcoin price has been explored as far back as 2011.

It has been commonly accepted that there is a relationship between the two but the exact nature of the relationship has always been murky. A deeper dive into the data by crypto data start-up Messari has discovered that the relationship between the price and Google searches for “Buy Bitcoin” is not as straightforward as it may appear.

Relating variables to price — the holy grail of investing

Breaking down how variables impact price is a practice which far outlives cryptocurrencies. Since the genesis of the equity markets in the early 1600s, the general public and professionals alike have been trying to break down the relationship between the price of an instrument its relationship to other variables.

It would make sense that a variable such as the Google Trends data for “Buy Bitcoin” searches would relate to the price. The more people that are searching “Buy Bitcoin”, the more people are likely to be buying bitcoin, the more buying pressure there is, and the price should increase to facilitate this.

Clear and logical. But the data shows it isn’t that straightforward.

What does the data say?

Messari analyst Qiao Wang (@QWQiao) examined the past five years of data for both Google Trends and Bitcoin price to explore how they relate to one another. Wang took weekly data points for both of these variables and calculated the weekly changes in the data to compare.

Wang analyzed how the “Buy Bitcoin” data relates to the price data at the same time but he also went a step further.

Wang analyzed how the price data relates to the trends data from the week beforehand. Wang also analyzed how the price data relates to the trends data from the week after.

This is a key part of the exploration of the relationship. It delves into leading-lagging relationships.

Does “Buy Bitcoin” trends data now have any predictive power to say what Bitcoin price is going to be a week into the future?

Does Bitcoin price data now have any predictive power to say what “Buy Bitcoin” price data is going to be a week into the future?

It turns out that the correlation (think association) of Bitcoin price now and Google trends data a week into the future is higher than the other relationships explored. Of the three relationships explored, the search trends of “Buy Bitcoin” had the lowest relationship with the price next week.

This makes sense. Price increases naturally drive the interest in bitcoin and users to search how to buy it.

It is FOMO materialized into numbers.

But is it reliable?

It’s great to show numbers that represent relationships. But it doesn’t really say a whole lot about what you can do with the relationship.

Wang went a step further with his analysis to see whether these relationships were “statistically significant”. This essentially means that there is less than 5% probability that the relationships observed occurred due to chance.

While all of the three relationships examined recorded positive correlations, only the relationship between current price data and trends data one week into the future was statistically significant. This indicates that there is a high probability that price increases today will drive higher Google searches for “Buy Bitcoin” one week into the future.

The other relationships were also positive but could simply have been due to chance.

This is all very enlightening. However, one BitcoinTalk managed to intuitively grasp this relationship back in 2011.

“There is a very nice correlation. But if you look closely you will find the best correlation peak where Google Trends lags the daily high by a few days. It is new highs that tend to drive interest, and interest tends to droop when the exchange rate drops. There are exceptions, but if you look at the entire time period over which Google Trends is reported for “bitcoin” you will find that this is the overall case… “

BitcoinTalk User Chodpaba on September 8th 2011

Now we have the statistical tests to back it up.

--

--