A Neural Networks Trading System

  • Client: Clever Financial and Financial Intelligent Systems
  • Date completed: 2006
  • Client’s goal: develop an automated trading agent mimicking a successful trader’s choice of stocks and timing regarding when to buy and sell them.
  • My contribution: an automated trading agent using a system of neural networks was developed, which was focused on the stocks comprising the S&P 500 US stock index.  Such networks were trained with a series of successful trades the trader performed, along with a series of technical indicators describing the context of the trades.
  • Results: These networks were tested using real data from the same stocks, and its results were compared versus the same index. Such results can be appreciated on the chart to the right.
  • Note: this project was part of my thesis to receive the bachelor’s degree in Computer Science. Even though the performance of the system was not impressive, this project opened a field of research regarding the mimicking of human trading though artificial intelligence. Hence a new book I co-authored and based on my thesis was published (so far available only in Spanish):

Performance of the developed strategy versus the S&P500 Index

Leave a Reply