A Multi-Channel Machine Learning system, method and computer program is for automated simultaneous transmission of a number of Buy/Sell orders generated according to differently self-optimized trading parameters for each independent trading strategy. Optimization could be performed according to every optimization method used. Perpetual real-time optimization (or self-optimization) of trading parameters adds Machine Learning feature to the invention and it is done independently for each trading strategy. A centralized trading system is provided for the individual user or organization that wants to perform his trading automatically and completely without human intervention from receiving the data from the central server provider to real-time order execution in computerized financial markets. The disclosed system is working based on an API/DSK programming tools and is integrated into a network of brokers, banks and other institutions trading on computerized markets, through a number of parallel working connection channels.