|  OCT 1994

Trading The S&P With A Neural Network by Jeremy G. Konstenius

Trading The S&P With A Neural Network by Jeremy G. Konstenius Here's an example of the steps that one trader took to use a neural network to trade the stock market index futures. Today's technical traders have the opportunity to use sophisticated methods for analyzing and developing trading strategies. More and more traders are looking at artificial intelligence, specifically at the use of neural networks to develop new trading strategies such as forecasting market movements. Artificial neural networks are mathematical simulations of biological neural networks, in which a neural net is shown data and instructed to respond in a particular way. The input data is the stimulus and the output is the response. The neural network learns the interrelationships between the stimulus and response by organizing itself internally. Once a neural network has been trained, it can be shown new data and will, hopefully, produce accurate responses. One popular model in the financial community is the back-propagation network. Back-propagation networks consist of artificial neurons that are interconnected to make up the network. These neurons consists of artificial synapses, which are numbers representative of intranetwork connection strengths. These connection strength numbers are modified during training by the back-propagation of error. If the neural network fails to accurately produce the desired results, the connection strengths are modified until the errors in the results are made tolerable by a trial and error process. The training process continues until the collective connection strengths of the neurons are modeling the input-output relationships with a certain degree of accuracy.

by Jeremy G. Konstenius