Berkeley Data Scientists successfully predict Bitcoin price!

The 2 things AI can tell you in the 2020 crypto bull market

For the second year in a row, AnChain.AI partnered with the Berkeley Data-X course to offer mentorship and support in leveraging the latest mathematical techniques and tools to solve highly relevant industry problems with the most impact.

This year’s data science innovation project aimed at finding signals that could provide insights into Bitcoin price movements. We found those, and went one step further!

The AnChain.AI Bitcoin Prediction team, who made it to the top 3 projects of the entire class, created a Bitcoin dashboard that forecasts the price, displays real-time trading signals, and shows live sentiment of users on Twitter.

Here are the 2 most important things AI told us about the 2020 crypto bull market:

  • Market sentiment plays a major role in influencing the behavior of the cryptocurrency market
  • Deep learning techniques can be effectively used to predict price fluctuations

Let’s dive deeper into the details of the project.

Harness the power of deep learning and make smarter Bitcoin investments

Is it even possible for non-professional traders to consistently profit from the highly volatile Bitcoin market?

This Fall, a group of UC Berkeley students partnered with AnChain.AI to create a product called BTC Predictor. BTC Predictor is a dashboard that features a real-time Bitcoin exchange rate predictor, day-trading signal gauge, and a Twitter sentiment chart that empowers day-traders to make more reliable investment decisions while minimizing their risks.

“Cryptocurrencies have historically been characterized by great volatility, making them a risky bet for most investors. This tool brings the power of data and predictive analytics into the equation, making the investment in digital assets less daunting.”

— Francesco Piccoli, AnChain.AI Product Manager

Image for post
Image for post
BTC Predictor Dashboard Demo

BTC Predictor is capable of predicting Bitcoin’s exchange rate in the next minute to a highly-accurate degree and in real time with little delay. There is no similar software that can make such predictions on a minute-to-minute cycle. The underlying model, thanks to an artificial Recurrent Neural Network (RNN) architecture, accounts for human irrationality by performing sentiment analysis on recent Twitter data and is also capable of reading multiple technical indicators at the same time. This allows it to capture market trends that would be humanly impossible to analyze, making the tool ideal for investors of any experience level.

Every minute, the Twitter tool examines and analyzes tweets containing the word “bitcoin”. It then uses an AI technique known as sentiment analysis to analyze a tweet’s polarity (how positive or negative the text is) and subjectivity (how factual or opinionated the text is). You no longer have to waste your time scouring through #bitcoin tweets to extract useful information, it’s already done for you! You can simply observe the charts to know how people around the world feel about Bitcoin at this moment.

Image for post
Image for post
Twitter polarity (in red) against BTC daily open price. You can easily see how people’s sentiment on Twitter anticipated the bull run of the past few months.

The technical analysis tool takes the guesswork out of the Bitcoin trading process. It provides users with a real-time trading signal in the form of a gauge that ranges from strong sell to strong buy. The tool was created using 12 trading strategies: 10 EMA crossover strategies and 2 Ichimoku Cloud-based strategies. Each of these 12 strategies outputs a buy or sell signal, and the average trading signal is displayed on the dashboard, ready to support your day-trading decisions!

Ever since its release, Bitcoin has become one of the biggest players in the cryptocurrency market. Many people have attempted to predict Bitcoin exchange behavior, but it has proven to be a difficult task. One reason for this is Bitcoin’s extreme volatility. As Figure 1 displays, Bitcoin’s exchange rate fluctuates wildly in comparison to traditional financial assets such as gold and stock indices. Subsequently, it is difficult to model the exchange rate of Bitcoin using simple regression models, as they fail to capture such variance.

Additionally, unlike the equity market where companies’ performance metrics are the mainstay of investor confidence, market sentiment plays a major role in influencing the behavior of the cryptocurrency market. Analyzing this feature adds another layer of difficulty to the already complicated prediction problem.

Image for post
Image for post
Figure 1: Historical Volatility of Bitcoin, Gold, HS300, and SP500 over 30 Days ¹

While there are several well-known models that have attempted to make Bitcoin exchange rate predictions, many of them hinge on flawed comparisons and assumptions. BTC Predictor avoids such pitfalls by retraining the underlying model with new data every 5–10 minutes. As a result, the model can anticipate shocks to the market and its predictions will not degrade over time as it constantly learns emerging market trends and sentiments. This makes BTC Predictor a reliable tool to boost your trading game and make smarter investing decisions.

BTC Predictor will soon be added to QTF, AnChain.AI’s new analytics product.

In the meantime, you can:

About the Berkeley DataX Team

The team consists of Chris De Leon (Computer Science), Michela Burns (Data Science), Vivian Lin (Data Science/Economics), Haoliang Jiang (Data Science/Economics), and was mentored by Francesco Piccoli, AnChain.AI Product Manager.

Disclaimer

This content and any tools or model referenced is not financial advice and it is not a recommendation to buy or sell any cryptocurrency or engage in any trading or other activities. You must not rely on this content for any financial decisions. Acquiring, trading, and otherwise transacting with cryptocurrency involves significant risks. We strongly advise readers to conduct their own independent research before engaging in any such activities.

References

  1. Li, Y. (2020, July 16). Enhancing Bitcoin Price Fluctuation Prediction Using Attentive LSTM and Embedding Network. Retrieved from https://www.mdpi.com/2076-3417/10/14/4872/htm

Blockchain data analytics firm providing security, risk, and compliance solutions.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store