Soudeh Sheybanifar
Abstract
Since noise present in financial series, often as a result of existence of fraudulent transactions, arbitrage and other factors, causes noise in financial data therefore false estimation of the parameters and hence distorts portfolio allocation strategy, in this paper wavelet transform is used for noise ...
Read More
Since noise present in financial series, often as a result of existence of fraudulent transactions, arbitrage and other factors, causes noise in financial data therefore false estimation of the parameters and hence distorts portfolio allocation strategy, in this paper wavelet transform is used for noise reduction in mean-variance portfolio theory. I apply conditional estimation of the mean and variance of returns along with the simple one obtaining “optimal weights” which later combines with smooth and non-smooth series, result in four optimal portfolio weights and therefore four portfolio returns. After this, I impose the non-negativity constraint (for weights) deduced from the Kuhn-Tucker approach to simulate the no short selling circumstance in Tehran Stock Exchange. Weights and portfolio returns changed dramatically in this step but the main result (which asset to hold) did not. Comparing Sharp ratios, I observed that Regardless of the psychological characteristics of the investor, holding the risk-free asset is almost the optimal choice in this case.
