Can the Price of BTC Bitcoin Be Forecast Successfully with NARX Neural Networks?

The cryptocurrency market data is highly noisy and nonlinear and conditioning the effectiveness and efficiency of the models that can be used to forecast their price, making this task a challenge, as shows the technical literature. Therefore, the present study proposes the following alternative: to train, test, and validate a NARX Open Loop Neural Network using the preprocessed historical data (Phase 1); then, to use a step by step technique with a NARX Closed Loop neural network to forecast in time (Phase 2). According to the results obtained, the forecasting of the prices of BTC Bitcoin is promising using data from the years 2017–2021, a period of high price variation. Finally, the subjacent idea in this study is that if appropriately modeled, BTC Bitcoin prices can be predicted reasonably accurately, although the cryptocurrency market is too hard to model.