Forecasting stockmarket timeseries data in a distributed framework

OData support
Supervisor:
Kazi Sándor Antal
Department of Telecommunications and Media Informatics

Trading stock exchange listed instruments such as stocks, bonds, derivatives makes inavitable to know how to aproximate the the price behavior of one lone or the basket of some given instruments. Since the pricing process of the stock excange is very complicated, dipendent of several factors, it could be required to precess and analyze a huge amount of data to achive this goal.

The object of the thesis to create such elementary models that can make approximations for the future instrument behaviors using the price data of the past. It was also an objective to build the models in such a way so they can make the researches effectively even on huge data sets, so the project was prepared in Apache Spark distributed calculation environment.

In order be able to make more accurate analysises I have created several models and have compared the results with benchark values.

In the first half of the thesis I will show the used technical methods with regards to the mathematical, IT, data analyzing and economical aspects of the topic.

In the second half I will show in detail the carried out researches and make the conclusions.

Downloads

Please sign in to download the files of this thesis.