Exploratory analysis of Crypto Twitter

In this post we will be exploring our Crypto Twitter (CT) dataset, which consists of 1.7M tweets from top ~1000 “crypto influencers” from 2013-2020. This dataset is strongly biased, since by definition the most popular Twitter users are the ones who write positive content about crypto.


Number of tweets

It seems that crypto influencers are tweeting about Bitcoin and Ethereum consistently, with a minor bump at 2017 year end crypto bubble. Tether on the other hand gets intermittent attention spikes, likely when some drama occurs (ie. lawsuits, investigations, etc.).

(Retail) Audience response

Unlike the tweets output from the influencers, their audiences response is much more tied to the price of the underlying assets. When the prices of BTC/ETH are at highs, the retail engagement trough likes/retweets/comments is high as well. Likewise, in bear markets it dries up.

Crypto returns vs. retweets/replies

If we difference the series, we can see that in general the CT audience is more likely to retweet or engage in conversations during months with exceptional returns. Monthly returns and changes in retail engagement exhibit a positive linear relationship on the log-log scale.


Top Mentions

Top Hashtags

Top Cashtags

Bag of words

Here we can see the most commonly used words, as a summary of memetics. gold, value, money, payments, government, HODL, Satoshi, price, speculation and altcoins are among the common keywords.

The Ethereum model shows a similar pattern, with addition of ICO, DeFI, tokens, Vitalik etc.


Lets look at most popular Bitcoin tweets during days of extreme volatility.

For these two examples I’ve chosen the second peak @ 14k on 26th June 2019 (although 20k top from December 2017 is a comedy gold mine as well) and the “COVID-19 sellof” of March 12th 2020.

Interestingly, most tweets on March 12th have a “positive” or “dismissive” attitude despite one of the largest historic % losses in USD value. This highlights the selection bias as mentioned in the introduction.

Top tweets during uptrend

Top tweets during market crash