Applying Technical Analysis of Stock Trends to Trading Strategy of Dynamic Portfolio Insurance

Abstract

In the trading operation of dynamic portfolio insurance, TIPP (Time Invariant Portfolio Protection), when adjusting active assets, only considers the scale of asset of that time regardless of how market trend proceeds. In other words, TIPP is clumsy in evading loss and pursuing profits. This study makes use of the predictability of artificial neural network, via market trend analysis and the learning of historical data, to find out the most optimized Multiplier of TIPP in various situations so as to optimize dynamic portfolio insurance. This study utilizes two kinds of artificial neural networks. One is to employ the price, quantity, and tendency technical index as the input item to predict the future rise or drop as the output item. The other is to employ the various technical indexes when MACD crossed on that day to serve as the input item, and the output items are the future range and days of rise and drop. The statistics show that the profitability of the prediction module of crossed MACD is better than the artificial neural networks; both are better than the traditional strategy operation of TIPP. Keyword: neural network, genetic algorithm, portfolio insurance 1 Research motivation and objectives The liberalized financial market has given impetus to the prosperity of stock market. Financial commodities such as bonds, futures, and options can be considered as hedging tools. Therefore, making appropriate use of such tools to lower investment risk while sustaining stable profitability by operations such as portfolio insurance is an issue which worth in-depth discussion. The purpose of portfolio insurance is to control the risk of investment portfolio at a reasonable level and protect the investment from losses of the portfolio net value while making profits when the market trend goes up. Portfolio insurance is relatively a good investment strategy for conservative and risk-reluctant investors especially when the stock trend is veiled and obscure. This strategy is especially suitable for the large funds such as balanced fund, insurance fund, retirement fund, investment trust, and etc. In dynamic portfolio insurance strategies, time invariant portfolio protection (TIPP) adjusts the weights of active and reserve asset investments to make profit and guarantee principal. Considering of the trading cost, dynamic adjustment of the ratio of investment portfolio is not feasible in the traditional way. Adjustment of active assets is done by periodical review of the cushion. However, such measure considers the total asset value only. Taking active asset in stock investment as an example, the total asset shrinks in bear market based on the set investment ratio. Contrarily the cushion increases along with the total asset in bull market. In other words, TIPP has shortcoming in active reaction to the market fluctuation. Some parameters of TIPP model are usually set according to managers’ personal experiences, which may cause inconsistent performances in different cases. Hence this study attempts to find an optimized way of setting these parameters by a systematic methodology. By analyses of general index trend, forecasts are done and mapping operations are executed to expand profit and reduce loss. Namely if technical indices are applied with the forecast ability of neural networks, the most appropriate multiplier of TIPP under every condition can be found through historical data learning. Optimization of dynamic portfolio insurance is then achieved.