Estimation of stock prices by Artificial Neural Networks and testing of behavioral finance approaches on these estimates

Past stock price movements and the factors affecting these movements were examined using various artificial intelligence methods in this thesis. Virtual markets have been developed similar to real values according to those examinations. Virtual investors, who differ from each other according to certain characteristics, were developed and made investment decisions in virtual markets. The developed virtual investors were also enabled to invest in the markets of the past. As a result, it has been revealed that which type of investor is more successful in virtual and real markets. As a result of the study, it was seen that the most successful virtual investor model is the model that uses fundamental analysis, examines the data of the past six months, and only invests in BIST100 companies with a one-month maturity. It has been determined that virtual investors' sensitivity to loss does not make a significant difference in return on investment.