Crypto trading: how much can sentiment analysis help?

Aggregated sentiment data

Take a moment to think about what drives the price of a cryptocurrency. Is it retail investors? Is it cryptocurrency whales? Is it partnerships with large and established corporations? Not exactly. At the end of the day, what causes someone to buy or sell an asset depends on how they feel about it! Every single decision to buy or sell not only cryptocurrencies but stocks and precious metals boils down to that underlying factor: how someone (either or a person or entity) feels about that asset. 

Here at CryptoMood, sentiment is what we’re all about. Our data scientists have developed two unique measures to evaluate how people feel about over 3000 cryptocurrencies in both the news and social media. This yields a social sentiment indicator, and a news sentiment indicator.  In theory, the opinion of the news drives the opinion of people, and the opinion of people drives the price of a crypto. 

Of course there is some back and forth there, which is why these sentiment indicators don’t “predict” price per se. They can however forecast price action in much the same way as meteorologists do with the weather. That being said, sentiment indicators alone are not enough to create an accurate forecast. From time to time you have to actually look outside (look at other factors that are more evident) to get a sense of where the price might go. 

Profits over the course of roughly 1.5 years = 2x+

With this in mind, a few cryptocurrency traders have been using our sentiment indicators as an addition to their existing strategies. One of them in particular, Petr Zurek, took the time to speak with us in depth and show us the impressive results he has had by adding our sentiment indicators to his repertoire. Petr ran a backtest which started in January 2019 and concluded in May this year. He used a trading bot and he programmed it with the following rules:

  1. Do not enter a trade unless there is momentum (a consistent move in price in the same direction, positive/negative)
  2. If sentiment is negative, trade short
  3. If sentiment is positive, trade long
Entry and exit for a Bitcoin long trade with sentiment analysis

In short, he used sentiment as a “filter” in addition to price data. In comparison to his previous strategy, the bot made much fewer trades (50 vs. 160) but the average profit per trade was 500$. Perhaps the most impressive statistic was the large trades. Using our sentiment indicators, Petr’s bot was able to leverage the massive shakeout in March. During that period, a few trades provided returns of over 6000$ each! The total profit exceeded 25 000$ from a starting amount of 20 000$ – a yearly gain of over 100%. 

Entry and exit for a Bitcoin short with sentiment analysis

Although these gains were quite impressive, Petr noted that he wished the bot would have executed more traders. Both he and our developers are on the same page: the indicators still need to be refined and any excess “noise” needs to be removed, but this is a very promising start. He has noted he will try playing around with different time frames (15m vs. 1 hour etc.) and using the same strategy with other cryptocurrencies, primarily Litecoin and Ethereum. He also believes the sentiment indicators may work even better with some other altcoin. 

Comparison between trading with sentiment analysis vs without

As it so happens, we recently released the Alpha version of our mobile app which is now available for Android and iOS. It is free to use but you are limited to BTC data only, with the most recent 4 hours obscured. Premium members get access to the data for over 3000+ cryptocurrencies as well as a live feed up news and social media posts which mention the cryptocurrencies they are tracking. If you’re not a fan of mobile apps, you can check out our desktop terminal which is currently in beta development. If you find any big bugs or provide constructive feedback and ideas, reach out and let us know. You may just receive our premium membership for free!

You might also like