According to a study published by the Southern Methodist University in the United States, the volume of tweets and the Google Search Volume Index (SVI) are important indicators for predicting Bitcoin and broadcast prices.
In the study, analysts collected notes from Twitter that mentioned Bitcoin and broadcast. The same was done using the Google Trends tool. Based on the ideas of previous studies, the hypothesis was that the number of Twitter entries and their message (positive and negative) can influence cryptocurrency prices. The study found that the number of Twitter posts and search queries on Google is one of the first indicators that change before price jumps.
The role of market sentiment in technical or market analysis is to uncover people’s attitudes towards the entire market or an individual index (in this case BTC and ETH). Sentiment analysis theory is a technical analysis section that says price trends are ultimately a reflection of crowd psychology.
Consequently, theoretically, if it is possible to assess how positively or negatively people relate to a particular stock or cryptocurrency, one can estimate its price trajectory. Although in this particular study, the volume of Twitter entries, and not the mood of market participants, was recognized as a leading factor in influencing the price of cryptocurrencies. The absence of pronounced sentiments, which are the leading factor, has been theorized because of the large amount of “noise” on Twitter regarding cryptocurrencies compared to real conversations about them.
For example, researchers found that there are 21 million bots on Twitter that publish mostly factual information about prices, advertising, spam, etc. Regular users have little or no real discussion about how they relate to BTC or ETH. Another problem that researchers found on Twitter was that the mood was mostly positive, even when BTC and ETH prices were falling.
People who tweet about cryptocurrencies, even when prices fall, are interested in them not only in terms of investing, which shifts entries to the positive. Despite their findings, the researchers did not completely rule out a sentiment analysis using various modeling methods.
Methodology and Results
In the study, analysts used the open source code VADER (Valence Aware Dictionary and Sentiment Reasoner) to analyze the data on Twitter. Twitter data has been taken since 2014 using bitinfocharts.com. SVI data has been taken since 2004, and its scaling was proportional to the number of searches on all topics for the terms Bitcoin and Ethereum.
As for Google trends data, the study found that the price is closely related to the search for the keywords Bitcoin and Ethereum and that these bursts of search queries occurred before actual price increases were observed. Another strong correlation between Twitter and BTC price was also found.
Finally, using machine learning, Google Trends results and tweet data were also put into a linear model to check for positive correlations. Data was divided between the training model and testing at 80% and 20%.
Last summer, another group of scientists conducted a similar study. Researchers at the Stevens Institute of Technology in the United States under the guidance of Professor Dr. Feng Mai found that comments and posts in social networks can influence the price of Bitcoin.
Featured image via Pixabay