What Actually Moves the Prices of Cryptocurrencies?

The research below is an excerpt from our Q3 Quarterly Report in partnership with eToro! Read the full report here:

The TIE & eToro Q3 Quarterly Report

What Actually Moves the Prices of Cryptocurrencies?

On July 11, 2019, Donald Trump tweeted, “I am not a fan of Bitcoin and other Cryptocurrencies, which are not money, and whose value is highly volatile and based on thin air.” We set out to test whether his assertion was true, or if there were a clear set of patterns that predicted price movement on individual digital assets.

"The TIE’s Crypto SigDev™ (significant developments) platform ingests real-time information from over 1,500 primary sources including token issuer blogs and forums, SEC filings, Central Bank announcements, Chinese publications, court cases, global regulatory rulings, media sources, cryptocurrency exchange blogs and API’s, and more. SigDev uses AI to categorize that information (for example, SigDev can identify whether a Canadian regulatory ruling mentioning NEO is related to NEO the cryptocurrency or NEO the Canadian stock exchange). Next, tags are used to assess the significance of any piece of information. Specifically, important news is categorized into the following buckets of significant developments: 51% Attacks, Airdrops, Announcements, Employment Changes, Forks, Funding, Halving, Illicit Activity, Listings, M&A Activity, Partnerships, Regulatory, Staking Announcements, and Token Burns. Appendix 1 provides an explanation for each of these individual events."

Leveraging The TIE’s SigDev and sentiment data sets we set out to identify whether news moved the crypto market and what, if any, impact investor sentiment had following an important announcement.

Which Significant Developments Move the Market the Most?

Significant developments are news articles referring to specific updates to blockchain protocols, teams, or and major events (see the table below for complete descriptions). Certain significant developments (or “SigDevs”) have been observed to have an effect on price. Using novel, automatic news detection systems developed by The TIE, we aim to understand exactly how SigDevs affect the price of cryptocurrencies.

Looking at maximum price increases within 24 hours after different SigDev announcements, we see that certain announcements tend to procure more responses than others. These are ordered from left to right in order of increasing average positive effect on price. These results are intuitive: one would expect 51% attacks — a negative assault on a blockchain — to not typically have a positive effect on price, while getting listed on a new exchange (“Listing”) increases demand on a cryptocurrency and thus should increase value.

Listing and Partnership announcements have immediate, large, positive influences on price. We looked into whether those increases are sustainable. Specifically, we are looking at the maximum price 24 hours after each SigDev announcement compared to the average price in the following week. The data shows that the hype of a Listing or Partnership announcement tends to wear off within a week; however, Funding announcements and announcements about Mergers and Acquisitions have positive effects that often extend beyond a week!

Why do exchange listings have the largest potential impact on the price of a digital asset and why is there such large variance in the data set?

When a cryptocurrency gets listed on an exchange it immediately becomes available to a large set of potential users that may not have had any prior exposure to that particular asset. This is not a phenomenon limited to cryptocurrencies — in the equities market, companies IPO to gain exposure to a larger group of prospective investors that may not have previously had access in private market trading.

However, unlike in equity markets, cryptocurrencies may trade on hundreds of individual exchanges, and thus different listing events may have significantly different impacts on price. In the case of Bitcoin, which is available on nearly every cryptocurrency exchange, a listing on another crypto exchange likely would have negligible impact on the price of the asset.

However, when a newer asset that was previously listed on more illiquid markets gets placed on a large trading venue, like eToro, Coinbase, or Binance, the potential price impact is the greatest. This is because markets previously didn’t exist for that coin on any of the larger exchanges (as measured by users and trading volume).

Consider the case of Aragon (ANT). On August 12th and 13th, Aragon was listed on a number of large cryptocurrency exchanges, including OKEx, Binance, and Huobi. The token was first listed on OKEx when it rose from $4.34 to $5.38 in just under two hours, a gain of 24%. Less than 24 hours later, the token was listed on Huobi and Binance concurrently, and the price of Aragon soared to a high of $11.45, a 164% gain in under 24 hours.

ANT’s price has since fallen to below its pre-listing value, but it is clear that in the short-term increased exposure via listings is a positive for the price of cryptocurrencies.

In contrast, when a cryptocurrency previously offered on larger exchanges gets listed on a smaller trading venue, there is no major noticeable impact on price in the short run. Consider the case of Bitcoin Cash (BCH). BCH was already available on nearly all major trading venues, including eToro, before it was listed on Crypto.com’s exchange on March 31st. Following that listing, BCH saw moderate upwards price movement (+1%) in an hour, but finished down over the following 24 hours.

What are my chances?

Considering the short and long term price changes after all SigDev announcements in the past year (about 10,000 data points), we are able to generate a loose expectation of price movement after a specific announcement. The tables below show both the average percent returns at the 1 hour, 1 day, and 1 week timeframes, as well as the probability of having a positive return at all.

Average Price Change Following Every Significant Development

Average Price Change Following Every Significant Development

Probability of a Price Increase For Each SigDev

Probability of a Price Increase For Each SigDev

Mergers and Acquisitions are remarkable, showing a 90% chance of a positive return after a week, averaging 8.23% in returns. This outsized return is likely due to the fact that most token-related M&A news are tightly held secrets. Further, M&A news in the context of tokens are typically done to add further value to an ecosystem. For example, FTX’s acquisition of Blockfolio brought a significant number of additional users into the FTX ecosystem. In just over a week following the acquisition, the price of FTX surged by 33%.

Conversely, and unsurprisingly, 51 % Attacks show the lowest probabilities of yielding positive returns. They show an average loss of 3.32% after one week. 79.2% of the time, 51% attacks result in a token losing value over the next 7 days. Following a slew of three consecutive 51% attacks, Ethereum Classic fell by 36% in just over a month. For example, the third of ETC’s 51% attacks led to the token falling by 23% over the following week.

Mainnet Launch or Upgrade results in a positive price increase for the coin only 47% of the time. A mainnet launch means that a token is moving from a testing environment to a live environment, giving users first access to the live version. We believe that the negative price movement is likely due to a “buy the rumor and sell the news” mentality. Often investors already know that a mainnet launch or upgrade is on the horizon and will buy a token in anticipation of the event.

Unsurprisingly, token burns result in the decreasing of the supply of a token, so intuitively one would think that the deflationary pressure would be a positive in the longer term, but early research on this subject has not offered any conclusive proof. As this industry develops over time, we believe that it is more likely that gains for token burns will be more sustainable, all else equal.

Does the source of the material matter?

Let’s be honest — as much as people like to say they get their news from real news outlets, many of us rely on Twitter. But does news from news outlets affect price more than tweeted information? It depends. For the most part, we find no significant difference in price action after announcements, whether they come from tweets or news outlets. However, we noticed that Regulatory announcements tend to have a larger effect when they come from news outlets, while Airdrop announcements are more supported by Twitter activity.

Indeed, the data reveals that most of the time it does not make a difference to cryptocurrency price whether a SigDev announcement comes from Twitter or some other source. However, it does make sense for these two cases. Regulatory announcements are most trusted from official sources and Airdrops are most effective when people are widely informed about them.

Does Twitter matter?

The boxplots above raise an interesting question: what effect overall does Twitter, or social activity generally, have on price after a SigDev announcement?

The plot shows how tweet magnitudes about a specific cryptocurrency may change after a SigDev announcement (from news outlets only). We see generally there may be a relationship, but it is not extremely strong (correlation = 0.53). Perhaps Twitter activity only has an effect on certain SigDev announcements.

Zooming in on just Listing announcements, for example, shows that there is a much stronger correlation on this subset of the data. This implies the more tweets there are after a listing announcement, the larger the price increase will be, which makes sense. The more people spread the word about a new listing, the more who will think to buy.

We can see a similar trend with Airdrop announcements. This one is a bit more interesting because the relationship extends into negative territory. This implies that if an airdrop occurs and people do not tweet about it (i.e. lower Twitter activity than usual), then the price of the token will decrease. This may be because an airdrop is effectively like a free giveaway of a token, and if people don’t see a reason to keep this token (i.e. because people aren’t talking about it), then they are likely to sell it right away. However, if people are tweeting positive things about their freebie, others are going to want to hop on board!

This is exactly what we saw with UNI in September!

It can be tricky to untangle whether price is increasing due to Twitter activity going up, or if Twitter activity is increasing because the price is going up — a classic “the chicken or the egg” problem. However, it is interesting to look at data from all announcements.

It is not always the case that increases in one will lead to increases in the other. However, in the case of Listings, it appears that increases in tweets are very likely to correspond to increases in price. This is convincing evidence that tweet volume does actually have a unique effect on price after a Listing (and airdrop, and, to a small extent, partnership and staking), but not necessarily other types of announcements.

So, there is evidence that there may be an effect of tweet volume on price after certain announcements (but not so much on others). Does the content of the tweets matter? In other words, how does sentiment play a role in price change?

The TIE also uses advanced algorithms to calculate the overall “sentiment” every 300 milliseconds based on Twitter activity. This is a measure of whether people are saying positive or negative things about specific topics (cryptocurrencies). For the most part, correlations are weaker between sentiment and price change (maximum sentiment reached within 24 hours after the announcement). For example, here we show maximum sentiment after Listing news announcements versus maximum price change. We can see the line does not fit the data as well as it does in the example above (correlation 0.6 vs. 0.8 above).

As opposed to a linear fit, an exponential relationship fits this data almost perfectly! This implies that the higher sentiment goes after a Listing Announcement, the higher price will go. Again, one could argue whether the sentiment is increasing because the price is going up or vice versa, but we don’t necessarily see such a relationship in other announcements. So this convincing evidence that there is a real, exponential relationship between sentiment and price increase after a listing announcement. If you see a new listing announcement for a cryptocurrency you own, you may want to start writing some positive tweets about it!

Despite what Donald Trump may think, the price of a cryptocurrency does not come out of “thin air.” Rather, there are real, tangible driving forces that determine the price of a digital currency.

Appendix