Department of Computer Science and Creative Technologies Harvard Case Solution & Analysis

Department of Computer Science and Creative Technologies Case Study Solution

1.  Abstract

Stock price prediction is never easy because of its huge amount of data, and it can be easily influenced by investor confidence in the market; the relation of the stock market and the political situation makes this problem much harder to tackle.Many people have tried to achieve this by using different methods for decades. The use of artificial intelligence in stock prediction could be found in the 1990s. To compare and find out which method is the best to solve this problem, I am going to review different systems that have previously been tried out by various people.

2.  Over view of AI in stock trading

The uncertainty of stock movements often causes investors to lose all their money, and between these investors many of them may just have little or even no knowledge about how the stock market works and the risks behind it;very often they just listen to the advice provided by a stockbroker or investment adviser, and these stockbrokers and advisers may not mention the details of the risks involved as they will receive a commission if the customer decides to buy or sell the stock. In order to provide a bigger picture for the investor and review the risks of stock trading, a system should analyse the stock data and provide objective advice for the user before they decide to buy or sell the stock.

Stock market activities are usually non occasional and no person or algorithm has ever succeeded because it involves a lot of data and a wide range of factors which cannot be quantified, for example the confidence level of the investor to the stock, the economic environment of where the company is based, the political environment, and so on. However, many experts have tried to address this problem, and recently an expert tried his new system and achieved a result of 68% accuracy for 4 years; in the meantime, another company named Renaissance Technologies had already use an algorithm to suggest investments based on their clients’ goals and risks with a 71.8% average annual revenue, all of which shed light on the future of artificial intelligence.

Department of Computer Science and Creative Technologies Harvard Case Solution & Analysis

The core purpose of this project is to figure out what and which possible artificial intelligence to use in financial market and how well artificial intelligence can predict the stock market. In order to archive this, a machine-learning programme will be built to perform some data mining work to find out the secret pattern behind many stock data, to make predictions about future stock prices for users’ reference, and give some investment strategy and advice to the user.

This write-up report is based upon five parts, including an introduction, research, design and implementation, system evaluation, and conclusions. In the research chapter, the methods and system tried out by a former expert will be reviewed; moreover, the reason for choosing the Bayesian network and the rationale behind it will be discussed. In the design and implementation part how the Bayesian network was built and the composition of the algorithm will be discussed. In the system evaluation chapter, the result and the performance of the system will be analysed. Finally, the conclusion chapter will discuss the result of the system.

3.  Approach of Solving Problem

We first find and choose different approaches to solving the problem by others which will be reviewed focusing on the system architecture, the rationale behind it, the result they achieved, the advantages and disadvantages of the system, and finally any improvements. Next,a different system design approach will be discussed, followed by the reason;a Bayesian network and rapid development design approach have been chosen.

3.1 Approach of Single System

In this approach of research, a different approach using a single system will be discussed. A single system means that the system only makes use of one kind of artificial intelligent algorithm. These will make the entire system less complex and more possible to produce.

3.1.1 Approach of Expert System

The expert system emulates the decision-making abilities of a human expert. The system is made up of five major parts; the user interface, the inference engine, the knowledge base, the case base, and a tuning engine. The user interface is used to let the user enter and receive information from the system. The interference engine is used to recognize and interpret the pattern for easier understanding of the pattern. The knowledge is used to store all the rules of the system, which are all related to past stock performances. The case base is used to collect real invested results and fine tunes the rules to adapt the dynamic situation. Finally,the tuning engine adjusts the knowledge after analyzing recent market trends from the case base. They have quite good results at that time, about 72% of the average hit rate in a multiple test.However, the system has some limitations; the knowledge tuning method to adapt to a dynamic situation and the knowledge base gained from the previous few years can still be useful. However, it often fails to predict the stock market’s timing accurately in practice. To deal with this problem, new knowledge needs to be evaluated before using real investments, and even though new knowledge is generated from knowledge acquisition with high costs, after it is evaluated and added to the knowledge base, it might often lose its usefulness................

This is just a sample partial case solution. Please place the order on the website to order your own originally done case solution.

Other Similar Case Solutions like

Department of Computer Science and Creative Technologies

Share This