Team: 84
School: SANTA FE HIGH
Area of Science: computer science genetic algorithms
Interim: We are attempting to analyze the market to develop a smart investing program. The stock market is an excellent example of a chaotic system that cannot be solved by simple methods. In order to come close to predicting such systems one has to use guess work. Genetic algorithms help us come up with a rough guess of what is happening. Genetic algorithms use Darwin's idea of evolution to evolve the best answer. The use of genetic algorithms in the market was started in the late 1990's and s growing to this day. With our project we hope to find patterns in the market can be exploited by a. The program is a step in developing smart programs that can learn from given data and apply what they are shown.
We well use genetic algorithms to look for patterns in S&P 500 stocks. The program will then compile a decision tree based on the best performing algorithm for each stock. We well reward the best algorithm and delete the worst. We also plain to use mutation and crossover, which are common and have been tested in genetic algorithms studies. Are program well have the ability to make decisions based on previous movement of the stock. With this information the program generates a population of possible answers. Each answer is then looked at the best are rewarded well the worst are deleted. We will mutate and crossover once the selection period has occurred. This improves the chances of the program finding the best solution and focusing on a bad answer. The nodes of the tree well are randomly placed with different trading variables. Throughout the testing period nodes well be altered in order to find the best system. Our variable we well test the amount of days the program looks at the stock before investing. We hope to find whether looking at a short change or longer movement of a stock is more profitable.
We hope to develop a program that can take chaotic systems and understand them. It is very important that this field be expanded. Programs that can find the best solution for a problem that no human could ever understand. We hope to better understand the decision trees and their efficiency.
Are group is moving forward with thorough research. We are just beginning prepare programming. We are also looking for a mentor again.
Team Members:
Sam Olmsted
Patrick Kerwick
Sponsoring Teacher: Anita Gerlach