The limits of the possibilities of artificial intelligence (AI) are still far from being known. This has yet to be proven through a study conducted by researchers from the Indraprastha Institute of Information Technology (IIIT) Delhi who combined AI with traditional finance to predict crypto prices.
Make way for the crypto oracle
This study was recently accepted by a well-known academic journal. Indeed, Elsevier Information Science confirms the effectiveness of the method developed by Delhi researchers for predicting cryptocurrency prices. This research work was made possible by PhD student Shalini Sharma and her supervisor Dr Angshul Majumdar.
The conventional art of predicting the future of assets
There are generally two predominant methods of prices in the financial markets. One is based on the conventional methods of the 1970s. It is the Baum-Welch probabilistic approach, which makes it possible to predict the prices on the financial markets, but also the uncertainty in the prediction.
However, this approach shows its limits in the face of cryptocurrencies, because the information on the underlying events at the origin of the price variations is unavailable. This is where the modern approach adds value.
Based on AI, this approach is also called deep learning or deep learning. However, the latter does not require any underlying knowledge or information. As a result, it still does not cover the uncertainty variable relating to cryptos.
Read future crypto prices like on a crystal ball
The observation before this study is that there were no tailor-made methods integrating uncertainty for crypto price predictions.
Shalini Sharma and Angshul Majumdar pulled out their magic wands for a new approach. Indeed, this solves the problem of uncertainty of predictions associated with cryptos. On the available data, one cannot do better in terms of precision. Also, this approach beats all the cutting-edge methods available.
To better understand the method proposed by the Delhi researchers, it is necessary to clarify the notion of CVI. The CVI is none other than the volatility index of cryptos. For this purpose, it takes up the fluctuation of a crypto over time. For example, a stablecoin like DAI or USDC will have a lower index compared to cryptos like Dogecoin or Shiba Inu.
Thus, the uncertainty estimates obtained by the two researchers are correlated with the historical values of the CVI of the available cryptos. Consequently, the predictions made through this new method are effectively interpretable.
It is a real revolution offered by these two scientists. Indeed, the prediction methods available before their studies were mostly content to be modeled on conventional approaches to financial markets. However, cryptos, like the blockchain from which they derive, come out of established patterns. It is one more step in the digital revolution that AI offers us.
Receive a digest of news in the world of cryptocurrencies by subscribing to our new service ofdaily and weekly so you don’t miss any of the essential Tremplin.io!