deep neural networks with noise induction effect for cryptocurrencyprice prediction

Over the past decade, with the advent of blockchain technology, we arewitnessing a dramatic increase in the use of cryptocurrencies. however, investing inthe cryptocurrency market is risky due to the market’s erratic behavior and high pricevolatility. accordingly, the need to use an appropriate model for forecasting in riskcontrol and management is considered intelligent. motivated by the aforementionedissue, we propose a new approach based on a deep neural network focusing on thepattern of errors. the proposed approach is based on the random walk theory thatargues that price movements are not all that random and that predictable componentdoes indeed exist. this new approach, tries to improve the forecast results by modelingthe residual values and inducing their effect on the main forecasts. we used the longshort-term memory (lstm) as the main prediction model and vector auto regression(var) for noise prediction on three famous cryptocurrencies: bitcoin, ethereum, andbnb. the results show that the proposed approach has been able to improve forecasts.