A Novel Portfolio Optimization Based on Funds Allocation with Short Selling Using GNQTS Algorithm and Trend Ratio

• Main Authors: CHEN,CHENG-YING, 陳政穎
• Other Authors: CHOU,YAO-HSIN
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/78jus4

碩士 === 國立暨南國際大學 === 資訊工程學系 === 107 === When investing in the stock market, investors first encounter the stock selection problem. Therefore, selecting a potential combination of stocks is a problem worth investigating. One commonly-used risk indicator is the Sharpe ratio. However, it defies the logic of investors because even an uptrend portfolio has a high risk. Thus, this paper proposes a strategy to improve the Sharpe ration denoted the trend ratio where the daily expected return is the slope of the trend line and the risk is the difference between the trend line and the fund standardization. Moreover, we propose doing normal trading and short selling simultaneously to increase the profit and spread the risk. We use the trend ratio to find a stable uptrend portfolio for normal trading and a stable downtrend portfolio for short selling. As there is no limitation to the amount of stocks in a portfolio, and because MPT’s computation complexity grows exponentially when the number of stocks increases, we utilize the quantum-inspired tabu search algorithm improved by the quantum-not-gate (GNQTS), to find an optimal portfolio in a large solution space. Besides, we use the sliding window to overcome overfitting problem. In addition of using the nearest time period as the training period, we use the same time period from the last year as the training period and tested it in the same period in current year, as some stocks have an economic cycle. Using the trend ratio while doing normal trading and short selling, with the GNQTS and sliding window, the experiment results show a promising result in which the risk is spread effectively and the profit is maximized.