Study On Algorithmic Trading Strategies Based On The Invisible Transaction Costs

According to the assumptions of traditional investment theory, the orders can be quickly executed at no invisible costs in the financial market with perfect liquidity. In this case, the investors will mainly be concerned with the topic of how to construct the optimal portfolio under various market conditions. However, the realistic stock market is limited with liquidity. The investors have to bear transaction costs, especially the price impact cost, when they execute their orders. In order to reduce the price impact costs, the investors usually split the large orders into several small orders, and then choose the optimal moment to submit to the market by algorithmic trading. Nowadays, existing researches related to algorithmic trading have not analyzed the estimation and main factors of price impact costs in Chinese securities markets, and not considered the influences of the opportunity cost and other invisible transaction costs on algorithmic strategies, and not analyzed the portfolio selection from the view of algorithmic trading strategy. These issues are inevitable in practical applications and need further studies.Based on the current researches of algorithmic trading, this dissertation analyzes the transaction costs, the measuring methods of transaction costs, and the major influential factors in the trading process. Then, this dissertation studies how to construct the optimal algorithmic trading strategy under the objective of minimizing transaction costs. Lastly, this dissertation researches the effects of algorithmic trading strategies on portfolio selection. The contents contain three aspects as follows.Firstly, this dissertation analyzes all kinds of transaction costs in the trading process, develops theoretical evidence when using volume weight average price as a benchmark price in the course of estimating the price impact. Using the high-frequency data of Shenzhen securities market, this dissertation estimates the price impact cost and analyzes its related influential factors, and compares the differences of those influences at different trading periods. The results show that the order size is significantly correlated with the price impact in the case of middle-cap and small-cap stocks. The smaller the order size is, the smaller the price impact will be. Considering the large market value and trading volumes in the large-cap stocks, the investor’s orders might be traded hiding among a multitude of other orders, which leads to the insignificant correlation between the order size and the price impact in the case of large-cap stocks. In addition, compared with other trading periods, the effect of order size on the price impact was the smallest during the first trading period after the opening.In order to describe the phenomenon that some outstanding shares do not participate in transaction, this dissertation develops an indicator to measure the liquidity of those stocks. The indicator is calculated by the ratio of the turnover to the market capitalization of tradable shares. Using the high-frequency data of Chinese securities markets, this paper analyzes the factors influencing the price impact, including the new liquidity indicator. The results show that this indicator is significantly negative correlated with the price impact. The market capitalization of tradable shares is insignificantly correlated with the price impact. However, the large-cap stocks may have the worse liquidity if most shares are actually in the state of illiquidity. Therefore, at the end of the year, the funds should pay attention to those stocks with lower liquidity in order to avoid the unfavorable behaviors from other companies. Meanwhile, the supervision organization should also pay attention to the large-cap stocks with lower liquidity in order to avoid price manipulation.Secondly, considering the situation that only some of orders can be executed on time, this dissertation develops a method to estimate the opportunity cost, and establishes a new algorithmic trading strategy to minimize implicit trading cost by using the optimization theory and methods. Moreover, this paper gives an analytical solution in the special case that investor can predict the total trading volume. The results show that the performance of our optimal algorithmic trading strategy is better than VWAP strategy. This new algorithmic trading strategy can effectively reduce the trading cost, improve the investment return, and provide an important basis for selecting the optimal trading strategy.Lastly, this dissertation studies the influences of the algorithmic trading strategy of the portfolio selection problem, and then analyzes how the frequency of investment portfolio adjustment affect portfolio returns by using the historical trading data in Shenzhen stock market. The results show that it is not the best to frequently adjust investment portfolios. The investors should consider the factors such as the stock expected return, risk and transaction costs, etc. After the comparisons of the total profits with portfolios adjustment at the interval of one year, semi-annual, one season, and two months, this dissertation finds that investors can achieve the maximum return when they adjust their portfolios quarterly.