Finansal piyasalardaki yatırımcı davranışlarının incelenmesi

Portfolio selection problem considers the optimal allocation of an investor's wealth among several risky assets within a market according to some objectives and constraints. In the idealized Capital Asset Pricing Model (CAPM) world, every investor uses the same mean-variance model of Markowitz and chooses the same efficient fund. CAPM asserts that the efficient fund is identical with the market portfolio. However, due to the violations of the assumptions of the mean-variance model, the efficient fund may not be identical with the market portfolio. The motivation of this thesis is to find the portfolio selection model that represents investor behavior in hedge funds and US stock market. The portfolio selection models we consider in this study can be classified into two categories: (i) Portfolio selection with utility models, and (ii) Portfolio selection with prospect models. For the utility based approach, we consider the portfolio selection models in Çanakoğlu and Özekici (2010). They focus on multiperiod portfolio optimization problem for an investor with hyperbolic absolute risk aversion (HARA) utility. In this model, asset returns depend on a stochastic market depicted by a Markov chain. We follow the rolling-sample approach in order to evaluate the out-of-sample performance of the portfolio selection models across hedge funds and US stock market datasets and find the best model that represents the market portfolio for each dataset. In the framework of prospect theory, we extend Taşkıncan (2014) by considering prospect investors within a market consisting of multiple risky assets and one risk- free asset. In our analysis, we use the piecewise linear and piecewise exponential value functions under the assumption that asset returns have multivariate normal distribution. We test the multiple risky asset prospect models for the US stock market dataset and report the results.