Predicting Bitcoin price and volume using online sentiment

Sentiment analysis is very prevalent in algorithmic stock trading. With this research, I aimed to establish causal relationships between online sentiment and the Bitcoin markets. The findings can be used to improve trading model accuracies.

Team size

Individual

Roles

Research, code

Duration

2017, 4 months

Investigating multi-channel discourse

This research aims to identify how Bitcoin-related news publications and online discourse are expressed in Bitcoin exchange movements of price and volume. Being inherently digital, all Bitcoin-related fundamental data (from exchanges, as well as transactional data directly from the blockchain) is available online, something that is not true for traditional businesses or currencies traded on exchanges. This makes Bitcoin an interesting subject for such research, as it enables the mapping of sentiment to fundamental events that might otherwise be inaccessible.

By analyzing this data on sentiment and volume, we find weak to moderate correlations between forum, news, and Reddit sentiment and movements in price and volume from 1 to 5 days after the sentiment was expressed. A Granger causality test confirms the predictive causality of the sentiment on the daily percentage price and vol- ume movements, and at the same time underscores the predictive causality of market movements on sentiment expressions in online communities.

NEGATIVE NEWS
EXCHANGE PRICE MOVEMENT

This research surfaces an interesting pattern in the behavior of online Bitcoin communities. News and forum are seemingly used to collect trading intelligence, whereas Reddit seems contain discussion as to what happened on the markets. We also find that volume changes are leading indicators for negative and positive forum sentiment and negative Reddit sentiment. None of the analyzed channels show a predictive causality from negative or positive sen- timent to changes in trading volume.

EXCHANGE PRICE MOVEMENT
DISCUSSIONS ON REDDIT

If we investigate relations between market movements and online sentiment (e.g. the reverse), we find that for some channels market movements in fact precede discussion - which seems intuitive in the context of the other sentiment channels. many more of such relationships can be found in the paper below.

NEGATIVE NEWS
EXCHANGE PRICE MOVEMENT

For news and forum channels, we most notably find a Granger-causality from negative sentiment to the price change, indicating that negative news sentiment has value in predicting these movements. This observation is reversed in negative Reddit sentiment, where sentiment is Granger-caused by percentage changes in price.