Deep learning and technical analysis in cryptocurrency market

A large number of modern practices in financial forecasting rely on technical analysis, which involves several heuristics techniques price charts visual pattern recognition as well other indicators. In this study, we aim to investigate the potential use those information (candlestick indicators) inputs for machine learning models, especially state-of-the-art deep algorithms, generate trading signals. To properly address problem, empirical research is conducted applies methods 5 years Bitcoin hourly data from 2017 2022. From result our confirm strategies using approaches. We also find that among specifically recurrent neural networks, tend outperform others time-series prediction.