Trading strategy based classification on cryptocurrency price prediction

These days, the price prediction of cryptocurrencies, which is one of the main topics discussed in the media, is considered as a difficult problem when the volatility and high risk is taken into account. In this study, classification and regression algorithms have been developed and compared with the time series model on the question of whether it will increase more than different threshold values in the next 24 hours by using different indicators as support and resistance levels, highest low opening and closing prices or not. One of the most important results obtained is that the use of limited data to cover a certain period, rather than the use of all data, gives better metric results.