Artificial Intelligence driven Portfolio Management

In general, portfolio optimization is used to strategically allocate funds into financial assets with the goal of maximizing returns over risk. In the traditional sense, such procedures necessitate a great deal of experience and rely on a trial-and-error approach. To address such inherent issues, this study provides a computational framework for integrating Monte-Carlo simulation-driven stock price optimization coupled with deep learning-driven forecasting of future stock prices. The Monte-Carlo simulation (MCS) is utilized to quantify the uncertain variations in the optimal parameters such as Sharpe ratio and volatility. The time history of the stock prices corresponding to the best stock (based on high returns over minimal risk) identified by MCS is further used to train the LSTM (Long Short-Term Memory) deep learning model. To ensure sound prediction capability, the constructed LSTM model is validated with out-of-fold unknown temporal history data. The presented coupled framework resulted in a promising method of portfolio management.