Abstract:
In this thesis, we investigated the portfolio choice and trading characteristics of individuals using data from pension accounts. The data contains nearly 17000 individuals' accounts during 'the years 2008-2013. Various and seemingly inefficient patterns are observed. We observed that as the time flows, average equity allocation increases. Not-married people significantly hold more equity in their portfolios. Individuals who are middle aged(35-44) hold equity in their portfolio most. We could not find a difference in the equity a11ocation characteristics between different genders. As for trading habits, males and married people significantly trade more than their opposites. In addition, people who are between age 44-54 trade significantly more. Moreover we observed high serial correlation in portfolio change decisions. Briefly, equity allocation and trading decisions are affected by demographic characteristics, also there is serial correlation in that decisions. These habits constitute behavioral biases. Furthennore, we assess the performances obtained by the individual pension fund investors. Majority of the people in the plan do not take any risk, thus obtain low return but high Sharpe ratios. We also studied return predictability in an aim to improve the performance of the portfolios by employing a simple trading algorithm. In general, we found little predictability in short term using classical methods. We observed that as we increase the investment horizon, the predictability increases. We claim that individual investors who lack the financial intellectuality and have aforementioned behavioral biases, could use the help of automated algorithms to ameliorate the performance of their portfolios. v
