University of Technology, Arts and Design

Cryptocurrency Prediction

  • DESCRIPTION: The field of cryptocurrency analytics is not a new concept. As in any exchange market, as in the stock market, in order to make predictions, it is necessary to analyze historical trends and external variables that may cause the price value to fluctuate in the future. Successful prediction of these values ​​will maximize the investor’s earnings. This project proposes the use of LSTM networks fed with OHLC (Open-High-Low-Close) values ​​of the trading pair to generate models that predict the future closing value. LSTM networks are optimal for extracting relationships and trends in time series, thanks to their ability to “memorize” in the long term. To select the best hyperparameters that will minimize the loss function, a GridSearch will be performed.
  • CREATED: Jaime Ocón González
  • DEGREE: B.S. Computer Science