What do quantitative social media metrics say about the current crypto market?
Sentiment is increasingly recognized as a factor in financial markets. Arguably this is even more true for cryptocurrency markets than for traditional markets. Combined with the vast amount of data that is available nowadays this offers some interesting opportunities for looking at cryptocurrencies in a non-traditional way.
In order to gain insight in to this I created algorithms to explore quantitative big data sentiment analysis for cryptocurrencies, starting with Twitter as a source. These metrics a shared frequently in the newsletter for which one can subscribe at AltcoinAnalytics.com.
So let’s look in to the actual quantitative analysis. The graphs below show output of the algo’s along with an explanation of what they actually show. As with all new metrics, some will provide interesting information while others will not, so challenge yourself to form an opinion.
Is crypto exposure on Twitter rising or falling?
What am I looking at?
This shows all tweets with 150 largest crypto cashtags (eg $BTC, $ETH), multiplied by the number of followers of a tweet. Rather than simply looking at the number of tweets also the quantity of followers is taken in to account. This gives for example bots less weight, assuming that they have significantly less followers. Basically, it shows the reach of crypto related tweets. The simple average of each crypto is taken in order to give smaller crypto’s more weight in picking up a signal. Then as a final step two moving averages are shown in order to see if a trend is persisting.
Since the start of running the algo a declining trend is visible. It will be interesting to see what prices to when exposure starts rising again.
And what is the tone of all these Tweets?
What am I looking at?
This shows the average sentiment of tweets, again simple average by crypto. This algo looks at the tone of all tweets (positive or negative) and plots this over time. This is more volatile than exposure, as is expected since tone can change day by day. Here again a decreasing trend is visible, and an uptick would provide for a nice case study.
But are there crypto’s which have significantly more Twitter attention?
What am I looking at?
The total crpyto market is interesting, but which individual crypto’s stand out? In order to look in to that the algo looks at exposure and sentiment relative to size. The intuition is that $BTC would like get more exposure than for example $DROP. In order to correct for that the graph above shows the top 10 sentiment exposed crypto’s relative to its size. It picks up if a small cryptocurrency has much exposure relative to its size, so if a small crypto is reaching many people it rises in the ranking. This is subdivided in four categories:
- Most exposure (Tweets * followers) relative to market cap (top left)
- Most positive sentiment (sum of positive Tweets minus negative tweets) relative to size (top right)
- Most Tweets relative to market cap (bottom left)
- Most positive average sentiment (bottom right)
Measures are shown as a Z-score, which basically shows how much it is higher than the average.
What other quantitative metrics are being worked on?
Many. The research is still exploratory and therefore maybe even more interesting to be the first to have a look at. The algo’s provide many new data every day, and the list below gives some idea on what is being looked at in the future:
- Average sentiment on crypto news sites/blogs
- Statistical analysis on which crypto’s react most to social media sentiment (probably the smaller cap coins)
- Sentiment on forums
- Overview of which crypto’s show the largest increase or drop in sentiment compared to previous levels
