Ανίχνευση γεγονότων ανόδου / καθόδου σε limit-order-book δεδομένα αγορών με χρήση SVM ταξινομητών
An SVM-based classification approach for the detection of upward / downward events in limit-order-book market data
Limit order book data include information about a transaction that has been completed in the past. In this case, the transactions related to the cryptographic value known as Bitcoin. By using SVM and also by extension, a classification approach, these data are organized and offer future results that are very close to Bitcoin's actual future value. The practical part is applied to Matlab programming language. The results, graphical and numerical, show the possibility of a well-predicted event occurring or not occurring at the next moment. The aim is to find the optimal parameter’s selection and achieve the best time and quality result.