Year: 2006
Type: dissertation
Abstract: Stock trading is a number of stocks to be exchanged from one trader to another \ntrader. It consists of a trader selling a number of stocks at a price and a volume, \nand another trader buying the same stocks at the same price and the same volume. \nMost traders want to buy a stock at a low price and to sell the stock at a high \nprice in order to make a profit. However, it is difficult to know whether the \ncurrent trading (buy/sell) price is low or high. Some researchers have presented \ntechnical trading rules which are mathematical formulas with many parameters to \nsolve this problem, such as moving average rules, filter rules, support and \nresistance, channel break-out rules, and so on. All these rules are based on \nhistorical data to generate the best parameters and use the same parameters in \nfuture trading to make a profit. When the parameters of a trading rule are set \nproperly, the trading rule can help the traders to make a profit (buy/sell at a \nlow/high price). Experiments have shown that technical trading rules are \nprofitable. \nHowever, there are still some disadvantages and limitations to the technical \ntrading mies in real stock market trading. First, the technical trading rules do not \nintegrate domain knowledge (expert experiences and domain constraints, etc). \nFor example, some trading rules pattern maybe only generate three signals during \none year trading to get the most profit. However, the pattern is unreasonable and \nit is unprofitable in future trading, because the pattern is only a mathematical \nmaximum, but it is impracticable in stock trading. Second, the output of a \nparameter for a trading rule is only one single value. Sometimes, it may be a \nnoise so the trading rule is inapplicable in future trading. Third, present \nalgorithms to calculate parameters of trading rules are inefficient. Most trades are \nperformed through internet such that they can buy and sell stocks in online and a \ntrade is completed in a second. Real markets are dynamic such that trading rules \nhave to be updated all the time depending on changing situations (new data come \nin, new parameters will be recomputed). Current enumerate algorithms waste too \nmuch time to get new parameters. However, a one-second short delay in real \nstock trading will lose the best trading chance. Fourth, when we evaluate the \nperformance of a stock, we need not only to consider its performance (profit and \nreturn), but also to compare it to other stocks performance. At present, trading \nrules do not compare to the other stocks performance when they are selected to \ngenerate a signal, so the selected stocks or rules may be not the best ones. Fifth, \nin stock markets, there are many stocks and many trading rules. The problem is \nhow to match and rank stocks and rules to combine a profitable and applicable \npair. However, trading rules do not solve this problem. Lastly, trading rule \ntechniques do not consider the sizes of investments. However, in real market \ntrading, different investments will result in a different performance of a pair. \nWe propose in-depth data mining methodologies based on technical trading rules \nto overcome these disadvantages and limitations mentioned above. In this thesis, \nwe present the solutions to combat the existing six problems. \nTo address the first problem, we designed a domain knowledge database to store \ndomain knowledge (expert experience and domain constraints). During the \ncomputing procedure, we integrated domain knowledge and constraints. We \nobserved the output more reasonable as we considered domain knowledge. \nTo address the second problem, we optimized a sub-domain output instead of a \nsingle value, in the sub-domain all combinations of parameter can get a near-best \nresult. Moreover, in the sub-domain, some experienced traders can also set or \nmicro-tune parameters by themselves and a better performance is guaranteed. \nTo address the third problem, we adopted genetic algorithms and robust genetic \nalgorithms to improve the efficiency. Genetic algorithms and robust genetic \nalgorithms can get a near-optimal result in an endurable execution time, and the \nresult is also near to the best one. \nTo address the fourth problem, we applied fuzzy sets and multiple fitness \nfunctions to evaluate stocks. Because many factors influence the performance of \na stock, it is necessary to create a multiple fitness function for genetic algorithms \nand robust genetic algorithms. \nTo address the fifth problem, we built a stock-rule performance table to rank \nstock-rule pairs and find the best matching pairs. The stock-rule pair results \nshowed that the ranked performance is better than that of randomly matched \npairs. \nFinally, to address the sixth problem, we drew a graph of the relationship \nbetween investments and number of stock-rule pairs to search maximal points, \nand to decide the number of pairs for different sizes of investments. \nIn summary, the purpose of this thesis is to identify optimal methodologies in \nstock market trading, to make more profit with less risk for investors. The \nexperimental results showed that the methodologies are more profitable and \npredictable.
Author: Lin Li
Institution: University of Technology Sydney
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W2580397022
