Research: Google search data can predict the price of bitcoin

FinTech | October 19, 2015 By:

Academics from the University of Cagliari, Italy, compared USD trading volume with data from Google Trends in the 12 months leading up to July 2015.

The results, detailed in their new paper, indicated that search volumes for the keyword ‘bitcoin’ correlated with – and sometimes predicted – the currency’s market volumes.

Using a Pearson correlation, which shows the relationship between two variables on a scale of -1 (lowest) to +1 (highest), Martina Matta, Ilaria Lunesu and Michele Marchesi found a positive correlation equal to 0.6. This means trading volumes “[follow] the same direction pace of queries volumes,” they said.

The trio found the Google Trends data also functioned as a “good predictor”, because of its high cross correlation value (calculated using a Granger causality series). This means query volumes were able to anticipate trading volumes in “almost three days”.

The researchers say they would like to test out their hypothesis on social media platforms such as Twitter, Facebook and Google+.