Artificial intelligence (AI) has long been used to predict various events. Thus, algorithms, based on historical data, predict Oscar winners, stock quotes and even the spread of the pandemic.
Over the past five years, efforts have been made to predict the bitcoin exchange rate linked to the dollar. By the way, startups, research groups, and enthusiasts are always trying to create an algorithm. It could predict digital gold behavior on the exchange in the short and long term.
Recurrent neural networks and predictive analysis
Developers of machine learning algorithms use different methods to create predictive tools. Indeed, the most popular of these are the Regular Nuclear Network and the Long Term Memory Model (LSTM).
LSTM is a cyclic neural network variant that can remember long-term addiction. Similar to how we use past experience to predict future events. It is a neural network that can remember information over long periods.
The first attempts to forecast the price of bitcoin were made during the boom that occurred at the end of 2017.
The developer attempt fails. The algorithm only considers historical price data. Although the cost continues to be influenced by many other factors.
Other attempts to develop an algorithm
Another attempt comes in 2019 by the US developer of machine learning algorithms Marco Santos [Marco Santos]… Marco uses the LSTM model for his forecasting tool. It chooses Yahoo Finance as its data source. According to the developer idea, the algorithm was able to analyze price fluctuations over the past 30 days. And, it can also forecast bitcoin rate against the dollar for the next 10 days.
In 2020, Australian developer Derk Zomer [Derk Zomer] develops an algorithm that predicts the price of bitcoin for the next 20 minutes. In fact, it uses a circular neural network and an LSTM model. The results show that neural networks can predict the course. To do so, it relies entirely on simple financial data. However, according to the developer himself, from a business forecasting perspective, such a model is completely unnecessary.
Another example that deserves attention is the algorithm of the American entrepreneur Frederick Riveroll. [Federico Riveroll]who sought to predict the use of artificial intelligence using not only historical price data but also news headlines.
For its algorithm, Frédéric uses two datasets: Bitcoin versus USD and Fox Company News. The neural network identifies a number of patterns between keywords in news headlines and the price of the first cryptocurrency.
Well done, you’ve read this whole article!
To congratulate you, we will offer you our free method of investing in cryptocurrencies simply and automatically with the aim of generating a regular income while protecting your capital as much as possible.
To access your gift, click here and register: https://www.fructify.io/actucrypto