Daneel's AI & sentiment Analysis features will soon become available on Cryptohopper

On Monday we were proud to announce that we just launched a collaboration with an AI app called Daneel.io. Daneel have succeeded in making an app that can scan the market sentiment for cryptocurrencies through scanning thousands of discussions, news sites and articles. In order to truly understand why this is an important development for us, it’s first important to explain what sentiment analysis (Aka opinion mining) is. In this blog, we'll go over sentiment analysis and show it's value for traders. 

Drawing Analysis from Crowd Psychology 

Sentiment analysis is one of the three forms of analysis in a traders toolbox, next to technical analysis and fundamental analysis. 

Sentiment Analysis uses technology to decipher public perception at a given time by drawing underlying opinions from a wide range of sources.

Recent breakthroughs in machine learning have advanced our ability to make these assessments accurately at mind-boggling speeds. Now, we are not just able to understand sentiments, but more importantly, we can track them in real-time.

Advancements in deep learning alone did not, however, make sentiment analysis the hot topic it is today. The quintessential part of the puzzle is the device from which you are reading this blog post.

With more than a third of the world’s population now connected online, sentiments on virtually all of aspects’ of life are reflected in comments, status updates, and headlines on our social media sites.

 Twitter, in particular, lends itself very well for extracting suitable data sets and has, therefore, become a gold mine for researchers in the field.

Some of its key advantages are the restricted length of the text, the ability to add descriptive hashtags and providing some of the fastest and earliest news updates online. Using AI, you can sort opinionated tweets by positivity, neutrality or negativity. Especially in the marketing and financial analysis industries,  this bears immense potential for getting real insights and foreseeing buying behavior. Understanding sentiments makes market research far more efficient and helps you sense trends the minute they materialize.

Tweet your heart out

Having emerged in the age of social media and strongly relying on global hyper-connectivity to function, cryptocurrencies and the internet go hand in hand. Sites like Twitter, Reddit and Telegram are popular places for tech-oriented crypto-enthusiasts to disseminate relevant information and share their thoughts. At the same time, as pure free market commodities, cryptocurrencies remain a highly speculative and volatile asset class.

As a result, public sentiment and valuation are inherently linked in the cryptocurrency industry. Unsurprisingly, social media posts describing cryptocurrencies are very promising targets for deep learning analysis, which have already demonstrated the relationships between market sentiments and price movements. 

The ways to conduct sentiment analysis 

To use sentiment analysis, you have two options.

  1. Choose a sentiment index (a sentiment tool)
  2. Create your own indicators directly from the data of exchanges.

In either case, the core principle for the analysis is the assumption that at extreme sentiment levels, the majority of interested market participants have either already sold (bottom) or bought (top), indicating that a reversal is imminent. The reason why this happens is that most people are risk-averse, meaning the number of buyers will decrease once profits can be made and vice versa.

However, this can become difficult to assess especially in very bullish conditions, where there is a steady expansion in the market. That is why some consider the technique to be more useful for identifying price bottoms rather than price tops.

Sentiment indexes that may yield positive results, could be rather simple websites like the “Crypto Fear and Greed Index.”   They measure and weigh dimensions like dominance, volatility, market momentum, and social media. Alternatively, you could subscribe to paid services, which have far more advanced machine learning algorithms and can study numerous different social media sites in parallel. 

You can also assess sentiments by reviewing data directly from exchanges.

There are two techniques you might find interesting for your analysis, namely looking at “long versus short ratios” and keeping an eye on “margin rates”.

By studying the long versus short positions ratio of a cryptocurrency, you are studying the percentage of market participants who are going long (expect the price to rise) or short (expect the price to fall) at a certain time.

A rule of thumb traders may go by could be that if 60% of all investors on the exchange want to short a coin, this would be indicative of a bearish sentiment, as the coin appears to be oversold and an upward correction is due. The reverse could be true if the majority of total trades represented long positions.

You can do a lot more research to understand how to interpret these percentage changes in different market conditions and find the ideal time to make a move. Potential sources for long versus short ratio information could be found on datamish.com or Bitmex.

Finally you can also use a more sophisticated tool like Daneel, our first official partner. They just launched a complete redesign of their flagship application to analyze the blockchain market sentiment even more in depth and with an improved interface and announced the web version to be released this week. This version, adapted for more advanced users, allows Daneel to be used on a computer or tablet with a higher level of data analysis to meet the needs of professionals. The analysis of the sentiment is now more precise and configurable: a doughnut graph of the Bitcoin sentiment and a visualization of the volume of the week's feeling will allow the user to go into details. In the future, they're planning the transition of their website into an automated news feed and an overview of the application's features to give an overview of the services offered. Through this, you can already view sentiment scores on all devices to see exactly what the overall opinions are of the coins you're interested in.

conclusion 

Sentiment analysis is therefore a very interesting complementary tool to go along with your technical or fundamental analysis. As stated earlier, aggregating attitudes on social media alone is not reliable enough to guide all your investment decisions. Also studying sentiments from the buying or selling intent on exchanges will not consistently yield positive results.

Nevertheless, contemplating changes in sentiments provides you with smart intel on what is going on in the space and may help you identify buy or sell opportunities with additional confidence, no matter your trading style.

The accuracy of sentiment applications will most likely increase in the future as deep learning algorithms become more proficient at studying online posts. Given that cryptocurrency investors are a highly internet-centric bunch, applying this knowledge could be an asset.   

We hope you have found this blog entry engaging and feel more comfortable to apply sentiment analysis as part of your trading arsenal. As always: Happy hopping!