crypto
Bitcoin crypto has drawn significant delight from policy-makers, investors, scholars, and regulators because of its explosive price appreciation. The value of Bitcoin rose notably during the 12 months from $788 on December 17, 2024, to $19,650 a year later, with a rise of 2394 %. The ongoing debate regarding the nature of Bitcoin attempts to ascertain if the virtual currency is a financial asset or an exchange medium, calling for classifying the currencies as financial instruments and researching the commonalities they might share with other known financial products.
This paper adds to the literature examining the intraday relationship between Bitcoin and principal traditional currencies, aiming to test if there is the same reaction to news sentiment and to complement previous evidence on currency features to support the debate. Specifically, we analyze the effects of high-frequency unplanned news announcements about Forex and Bitcoin on Forex returns, volume, and, volatility, and whether Bitcoin responds in the same way.
We give a thorough analysis of Bitcoin with a sample term of nearly seven years of 15-minute data from 1st January 2012 to 1st November 2018. Employing Raven Pack News Analytics 4.0, we build a sentiment index for both currencies and Bitcoin and we investigate the impact of currency returns, volume, and volatility from news sentiment through an exogenous vector autoregressive model.
Background and related literature
First introduced in 2008, Bitcoin crypto is a digital currency, that is, an electronic cash system with no physical counter value, and is divisible to any degree. There is no single market or central authority, but instead, currencies like Bitcoin are decentralized and typified by a peer-to-peer network dispersed over well over fifteen thousand exchanges. Every transaction has to be approved by other users, or nodes, to be verified and written to the public ledger, i.e., the blockchain.
We gather data2 for Bitcoin over the sample duration January 1, 2012–November 1, 2018, from bitcoincharts.com, a site delivering transactional data for the majority of the world’s Bitcoin exchanges. While Pearson correlation coefficients for news sentiments are near zero which motivates our initial analysis, bitcoin charts, latest bitcoin markets, latest currencies updates, highest currency volume we execute some robustness tests to control for the occurrence of commonality in news sentiment and possible multicollinearity that taints the main results.
More Information
Natural language processing was embraced in this research to frame data-driven discourses in the crypto economy, the Bitcoin market in particular. By utilizing topic modeling, specifically Latent Dirichlet Allocation, text analysis of 4218 currency articles published during 2022-2025 from 60 nations in global news media unveiled major topics connected with currency within the global news media during this period. Empirical evidence based on this study is that through the corpora of global news articles, there were 18 major topics described in terms of the following categorical macro discourses: crypto associated crime, finance governance, economy, bitcoin crypto, markets.
Analysis indicates that the discourses identified can have had a ‘social signal’ impact on movements within the crypto-financial markets, specifically on the price volatility of Bitcoin. Findings indicate these particular discourses were found to damage Bitcoin’s market price, within 24 hours of the publication of the news articles. In addition, it was discovered by the study that in a few instances, news sources may have enhanced the volatility effect, and this includes geographical location relative to the latest broader market condition.
Current literature
The role of the media is one of providing information and sentiment to the financial crypto market. Bloomberg or Reuters, for instance, as serious financial media can influence the markets since investor action can be due to company news and events that receive high media attention. Additionally, social signal effect studies of the social media site Twitter have found that opinion polarization and exchange volume increases come before increasing Bitcoin prices, and that emotional valence comes before opinion polarization and increasing exchange volumes.
Mai et al.’s results granted that social media sentiment is a significant predictor in the valuation of Bitcoin but identified how not all social media messages have of equal influence on latest Bitcoin price. For instance, the authors demonstrated how social media’s influence on Bitcoin is motivated by the silent majority, the 95% of users who are less engaged and whose input constitutes less than 40% of overall messages.
Media reports reflecting changing optimism and pessimism can be accounted for by the fixed effects, along with article length, style of writing, and access to information for various journalists. What the media tells their public about its formation, and how the news media reports on currency and its analysis, are just as significant and impactful on the crypto-economy as Google search and social media Bitcoin statistics.
Data preparation
Once the crypto text had been gathered and aggregated, the text was pre-processed in Python with the help of the SpaCy, Gensim, and Pandas Python libraries. Pre-processing was a necessary step before performing the NLP on the text. The NLP process essentially involved four general steps to load the input data crypto text articles, to pre-process the data, to convert documents into bag-of-words vectors, and lastly to train the LDA model. SpaCy was utilized to tag and parse a sample document.
It was here that the statistical models and trained pipeline were used, allowing SpaCy to predict what tag or label would be most apt to use in the context. One of SpaCy’s trained modules contained binary data that was generated by exposing the corpus to sufficient examples such that it could make predictions that generalized over the language—i.e., a word that followed the in English was probably going to be a noun.
One of the pre-processing steps was to train the phraser that automatically extracted common phrase and multi-word expressions from a stream of sentences. This involved lemmatizing the text articles where the base forms of words were assigned using SpaCy, tokenizing the crypto text articles splitting the text into words and punctuation marks, etc., and calculating bigrams multi-word expressions or common phrases using Gensim.
The crypto crime discourse
Generally throughout the entire news corpora in the dataset, the crypto crime discourse can be defined as an overarching category that includes news media reports that are about currency-based crime. These are for instance news reports on scams, scandals, or exchange hacks. Newsworthiness via publicity from news reports of these stories with a criminal focus seems to be having an adverse price impact on Bitcoin have a look at the graphs and subsequent descriptions below. There were multiple news articles of this story within this research study corpora of news reports that had the effect of becoming part of the influential discourse in the field of crime.
This is revealed through the output from LDA as a subject. In particular, though, a news article in the media published on 24 February 2025 reported on this development, in which news was announced that a judge ordered the distribution of bank drafts worth $30 million related to the case of insolvency. Publication of this news article was accompanied by a −9% price movement in the latest top Bitcoin market.
Conclusion
In conclusion, the study drew upon existing research and learned from the analyses of Mai et al. of media and crypto Bitcoin price to develop a new understanding considering traditional news media and how it contributes to Bitcoin’s volatility. This research thus presented empirical evidence that the most prominent topics related to currency in the global news media during the 2022–2025 period were discussed in terms of the following categorical macro discourses: related crime, financial governance, and economy and markets.
LDA topic modeling was employed as a computational approach to discover and represent data-driven discourses of cryptocurrency in news media to make inferences regarding possible social signals or sentiment impacts this had on the latest crypto currency markets over the specified period. This adds to the existing body of research and knowledge regarding cryptocurrency, whereby previous research has identified a definite link between the role of the media and the market, i.e. popularly used terms and market movement.
Drawing on the earlier crypto research, this study offered concrete illustrations of news media coverage based on each discourse, i.e., the crime, crypto-governance, and latest crypto economy markets. In addition, it demonstrated potential relationships between discourse theme or sentiment and volatility in crypto prices. The study established that the crime, governance, and crypto-economy/markets discourse damaged Bitcoin prices across two years.