Analyser for technical analysis indicators using historical intraday values

Analyser for technical analysis indicators using historical intraday values

Abcouwer, F.C.A.
Janssen, M.
Savenberg, X.
Verboon, R.S.

De Weerdt, M.M. (mentor)

Electrical Engineering, Mathematics and Computer Science

Computer Science

Computer Science and Engineering

2013-07-03

Technical analysis is widely used among traders, financial professionals and financial institution as a tool to identify trading opportunities. However, the academic world does not share this point of view, since they claim the evidence to promote the usability of technical analysis is sparse and inconsistent. Noticing this impasse we decided that developing software to analyse this issue would be a way to facilitate research on the subject, and might provide a way to make an educated decision on the usability of technical analysis. The goal is to build an application that runs a strategy and shows the results of this strategy to the user. Over the course of this project, the project group has developed such a benchmark system, with development proceeding according to an incremental, agile development technique called Scrum. The largest challenge we faced was storing and processing our data. After investigating the options, we decided to use MapReduce with an implementation of Apache Hadoop and reduced both file storage size and processing time by implementing Apache Avro. The final system can be divided into three main sections: Server - On the server side, the Hadoop framework was used to facilitate the implementation of a scalable MapReduce approach to the problem of going through a large amount of stock market data and producing trades using an implementation of a technical analysis-inspired strategy. Client - The client side manages the visualisation aspects of the project, with the GUI having been built using the JavaFX framework. Metrics - A calculation layer which processes the trades that the simulation generated, and calculates all the metrics shown in the user interface. While working on the project, we worked hard to ensure high code quality and high levels of maintainability and extensibility of the system using a diverse set of tools. Testing was also an important aspect of development: using JUnit for Java, we wrote unit tests for various parts of our application. In addition, we wrote integration tests to test the proper functioning of larger parts of the system all at once. The application delivered in the end implemented all of the features listed as ‘must have’ and ‘should have’, except for one ’should have’ feature: algorithmic optimisation of strategies. The reason for this is that implementing this feature would have taken too much time that it would have detracted greatly from the overall quality of the final application, and we therefore decided not to include it. At the end of this project, we have developed an application accepts a strategy along with parameters defined by the users, processes it in a few minutes while showing its progress to the user, and displays customisable information about the results of the simulation in an intuitive user interface.

bachelor project
technical analysis

http://resolver.tudelft.nl/uuid:a5fcec34-958b-419d-bffe-ec1d780a0bc5

2013-07-05

Student theses

bachelor thesis

(c) 2013 Abcouwer, F.C.A.
Janssen, M.
Savenberg, X.
Verboon, R.S.