Algorithmic trading and portfolio optimization using deep learning and high-frequency market data

Algorithmic trading has become a dominating form of trading in equity and derivatives markets. A recent analysis of trading data from the U.S. equity market revealed that 70% of all trading activity is now handled by computers using various algorithmic strategies and trading systems. These strategies range from traditional execution, liquidity shaping and smart order routing strategies to more sophisticated high-frequency trading strategies that are often based on statistical arbitrage and market making principles. Autonomous in their performance, proprietary trading firms typically run high-frequency trading strategies on collocated servers. These high-performance trading infrastructure are expensive and involve a highly-technical staff that build custom software intrusions, write often complex algorithmic trading strategies in low level programming languages, and configure powerful computing networks and fast internet connectivity. Not surprisingly, many proprietary trading firms have generated astronomical profits from the skyrocketing growth in trading volume, particularly in equity options.