Heikin Ashi Candlesticks for Cryptocurrency Returns Clustering

Trading systems suggest several charting techniques allowing traders to better visualize the price movement of their assets. Heikin Ashi candlestick is one of the charting techniques that allows traders to visualize the price and trend movement of a cryptocurrency, which helps them in deciding the next move with their assets. Within this work, we used the KMeans clustering algorithm, to cluster Heikin Ashi candlesticks and logarithmic returns, to investigate the best number of classes to be used for logarithmic returns of a selected set of four cryptocurrencies retrieved from the Binance historical data center. We found that for the considered cryptocurrencies, the best k-values range from three to five. These results suggest the necessity of the clustering task before any classification task when dealing with cryptocurrency logarithmic returns forecasting, instead of an empirically predefined range of classes representing uptrend, downtrend, and no change in logarithmic returns.