Data mining based smart fund portfolio optimization system

Portfolio optimization concept is very popular in daily use. With minimum risk-maximum return approach, optimal portfolios were established different models. However, in advancing techonolgy and fast developing scientific age basic optimization models are outdated. Financial analysist's, whom are wasting quite time because of dealing with very large data sets, with the help of data mining concept started to complete optimizations faster and in safer way. By this means financial analysists instead of wasting time with worthless data sets, they can complete their analysis in a very short period. In these thesis, mutual funds portfolio data sets are optimized and tested with clustering method one of the data mining techniques and artificial neural network analysis. The aim of this thesis; the designed model is to create portfolios with higher return and lower risk than the existing portfolio of funds. As a result of the model test, it was concluded that the fund portfolios generated were successful, that is, the fund portfolio yield was higher and the risk was lower. The artificial neural network model was chosen for this study because not only the state of future data is unknown but also it was intended a smart model for future period with future data. It has been determined that the returns of the portfolios created with artificial neural networks are higher and their risks are lower. In the final part of the thesis the model is turned into smart system with drawing models workflow chart. With up-to-date dates and models, it can update itself. The system and flow chart are also allow for the operation to be based on automation with different datas. This optimal model which is created in this thesis will have importance for users (financial analysists, individual investers, institutional investors) and literature.