Main menu

Pages

Bitcoin meets Google Trends and Wikipedia

 Bitcoin meets Google Trends and Wikipedia: Quantifying the relationship between phenomena of the internet era

bitcoin trends

Introduction:

The emergence of the internet has completely changed the functioning of the real economy by allowing almost all internet users to exchange and share information at no cost. One of the fascinating phenomena is the emergence of digital currencies on the financial markets such as: BitCoin, LiteCoin, NameCoin, PPPoin, Ripple and Fri. 

Indeed, a digital currency is defined as an electronic alternative currency and without a physical form. It is not issued by a specific central bank or government. On this premise, digital currencies are, therefore, detached from the real economy (not from macroeconomic fundamentals such as GDP, the interest rate and the inflation rate). 

The digital currency market is, therefore, dominated by short-term investors, speculators and risk-takers. This means that the prices of digital currencies are motivated solely by the level of investor confidence, hence their sentiment becomes a crucial variable. 

In this article, Kristoufek studies the relationship between two main phenomena: digital currencies (namely Bitcoin) and search queries on Google Trends and Wikipedia (as two indicators explaining the interest and attention of investors).

Objective:

This research is supported by Mr. Kristoufek in 2013 in order to study the relationship between Bitcoin prices and search queries on Google Trends and Wikipedia.

Problem:

What is the effect of the attractiveness of Google and Wikipedia on the value of bitcoin?

Methodology:

Data:

The time series for Bitcoin on the Mt.Gox market have been available since 7/17/2010 with a higher frequency at 1 minute. However, the Bitcoin market remains illiquid for the first year of its existence. For this reason, Kristoufek  divides the sampling period into two subperiods: illiquid period and liquid period.

Even though the BitCoin market is a 24/24 and 7/7 market, Kristoufek uses an 8-hour trading day (as a reference of a liquid market). Therefore, he analyzes the series from May 1, 2011 until June 30, 2013. 
For Google Trends, Kristoufek uses weekly data and, as such, he obtained 113 observations in total. While for the Wikipedia variable, daily data are adopted so that it finds 788 observations.

Statistical tool:

To test the cointegrated relationships between variables, Kristoufek uses two Johansen tests (the trace and likelihood tests). First of all, he uses the Vector Autoregression (VAR) model to analyze the cointegration between the Bitcoin series and the Google Trends series. While he applies the error-corrected vector model (VECM) to examine the second relationship between Bitcoin and the Wikipedia series.

Results:

- Kristoufek found that the increased interest in Bitcoin leads to an increase in its price. That is, interest in the currency is growing, demand is also increasing, leading to price growth. However, as the price of Bitcoin increases, not only the interest of investors as well as the general public. 

- Kristoufek found that in the first 7 days (a trading week), an increase in prices causes an increasing positive reaction from daily views. After the first week the effect stabilizes, however the interest in Bitcoin does not return to the initial level. On the other hand, the author does not observe any significant effects from daily views on Bitcoin prices. The difference between Wikipedia and Google trends can be caused by the fact that these two engines are different and that the people who use them may have different behaviors. Nevertheless, Kristoufek believes that both engines provide interesting information about how Bitcoin works and the relationship between this digital currency and the general interest of the public. 

- In addition to estimating the cointegration relationship between the variables studied, Kristoufek is also interested in whether the price reaction on Google trends and Wikipedia is symmetrical, i.e. whether an increasing interest related to price increases has the same effect as an increasing interest related to price decreases. A crucial disadvantage, related to the measurement of interest using search queries on Google Trends or the number of daily views on Wikipedia, is the fact that it is difficult to distinguish between interest due to positive or negative events. To distinguish between positive and negative feedback, Kristoufek introduces a dummy variable equal to one (if the Bitcoin price is above its level) and zero otherwise. 

- Kristoufek notes that, practically, any reaction comes from positive feedback because there is not a reaction related to negative price movements in search queries. While for the daily views on Wikipedia, the researcher finds that the positive and negative feedbacks are practically symmetrical around the zero reaction. That is, the reaction of Bitcoin prices on Wikipedia is similar. This implies that the distinction between positive and negative feedback makes it possible to determine the exact reactions of Bitcoin prices.

Discussion&Conclusion:

Digital currency is a new type of economic tool with special properties. Probably the most important of these is that they have no underlying assets they are not issued by any government or central bank and they do not carry any interest or dividends. Despite these facts these currencies and Namely the Bitcoin currency has caught the public’s attention due to an unprecedented price surge and could make hundreds of percent in profits in just a few weeks or months. 

In this article we analyze the dynamic relationship between Bitcoin price and the currency's interest Search queries on Google Trends and frequency of visits on the Bitcoin Wikipedia page. In addition to a very strong correlation between the price level of digital currencies and the two internet engines we also found a strong causal relationship between price and search terms. What matters is that we The relationship was found to be bidirectional i.e not only does the search query affect the price but the price also affects the search query. 

This is in line with expectations for financial assets without fundamentals. Speculation and chasing trends clearly dominate Bitcoin price dynamics. Specifically we found that while the price was high (above trend) the increase in interest drove the price further up. On the other hand if the price is below its trend the growing interest pushes the price even deeper. This forms a suitable environment Quite frequent bubble behavior has indeed been observed for the Bitcoin currency.

 We believe that this paper will serve as a starting point for studying the dynamics of the statistical properties and busting behavior of digital currencies as they provide a Study the unique environment of purely speculative financial markets.

Comments