May 10, 2018 anay_PCL

AIEVE : A lesson to predict the future – Part 2

How AIEVE exploits Sentiment Analysis in Social Media for Cryptocurrency Forecasting

AIEVE is running on top of state-of-art Big Data, Machine Learning and Natural Language Processing technologies.

AIEVE can analyse the market to predict the trends of every major cryptocurrency. She is capable of managing any given portfolio with a variable level of risk as elected by the user; on release, AIEVE will initially be using a low-risk strategy. In this highly volatile market, only a very good AI engine can manage a portfolio correctly.

In the previous article, we showed the performance of AIEVE with a $1000 portfolio. Indeed, from January 14th to April 14th, she’s been able to increase the value of her savings by 88%. If you put the same money into any single coin instead of investing with AIEVE, you would have lost between 43% and 72% (depending on which major crypto asset you chose).

We had now improved this version. If you want to know how and see this in action, please continue reading and registering on AIEVE website here.

AIEVE is now able to analyse a large stream of social media data to better understand investor behaviour.

The role of Sentiment Analysis in Social Media

Twitter is a microblogging service that plays an increasingly important role as a social network and a news media.

The popularity of Twitter is producing a vast amount of short messages (tweets) that contains interesting information for Natural Language Processing tasks like Sentiment Analysis.

Sentiment Analysis is an active area of research that makes use of Natural Language Processing techniques to track positive or negative opinions in text in general or about a given entity, a product or an event.

Classifying tweets according to sentiment has many applications in political science, stock markets and cryptocurrency markets.

How important is Social Media for Cryptocurrency?

Twitter has become a critical social media tool that has an important effect on cryptocurrency market. A vast amount of messages related to cryptocurrencies appears constantly on Twitter with a possible immediate impact on cryptocurrency price.

The public mood expressed in social media plays an important role when it comes to decision-making. Individuals care about the opinion of others especially when they have a stake in a given cryptocurrency.

The high volume of social media messages necessitates automated processing to extract actionable information, which is not possible for an individual investor.

In the latest version, AIEVE supports Real-Time Twitter Sentiment Analysis.

How Real-Time Twitter Sentiment Analysis helps AIEVE?

It is well known that cryptocurrencies are highly volatile compared to traditional currencies. In this context, predicting cryptocurrency price can be very challenging.

AIEVE is now able to understand the relationship between the opinions expressed in social media messages in real-time and the likelihood of investors to buy or sell cryptocurrency.

AIEVE performance for 5 Cryptocurrencies

We evaluate AIEVE (with Twitter Sentiment Analysis and without) from 1st to 15th April 2018.

In table 1, AIEVE without Sentiment Analysis achieves a MAPE (Mean Absolute Percentage Error) of 0.2% by the average for 5 cryptocurrencies. Adding Sentiment Analysis improves forecasting effectiveness, AIEVE obtains an error rate of 0.15% and an improvement of 28% in comparison with the previous version.

Golem Project and Ethereum rise

As an example, from the 10th of April 2018 through 15th of April 2018, Ethereum’s price was up 21%. The main reason is the launch of Golem, one of the first Ethereum ICO projects.

Intuitively, if there is positive news about a particular cryptocurrency AIEVE will expect the price of the cryptocurrency to increase, whereas if there is negative information she will expect the price to decrease.

The figure below shows the rise of Ethereum’s price linked to a sample of positive tweets related to Ethereum and Golem project from the 10th April 2018.

During this period, AIEVE with Real-Time Twitter Sentiment Analysis performs really well with an error rate of less than 0.1% and an improvement of 60% in comparison with the previous version.

If you’re looking for more information or wish to follow AIEVE’s performance, please come to our website and register.

AIEVE will only work with PCL tokens, so be sure to buy now in order to get a good price!