na študijskem programu 2. stopnje Matematika
Anja Drevenšek (Author), Timotej Jagrič (Mentor), Marko Jakovac (Co-mentor)

Abstract

Umetna inteligenca in strojno učenje postajata del našega vsakdana. Tako je tudi na finančnih trgih, kjer t. i. inteligentni agenti trgujejo z vrednostnimi papirji. V magistrskem delu je obravnavan primer takšnega mehanskega trgovalnega sistema za trgovanje z delnicami. Model temelji na spodbujevanem učenju in uporablja realne prosto dostopne borzne podatke. V prvem delu so preučeni temeljni pojmi verjetnosti in umetne nevronske mreže. Nadalje je podrobneje opredeljeno spodbujevano učenje in matematično ozadje spodbujevanega učenja. Drugi del magistrske naloge predstavlja implementiran model mehanskega trgovalnega sistema. Zastavljeni in učeni so štirje agenti, ki se med seboj razlikujejo po sistemu nagrajevanja. Agenti so testirani in primerjani s pasivno strategijo ter med seboj.

Keywords

magistrska dela;strojno učenje;spodbujevano učenje;umetne nevronske mreže;inteligentni agent;trgovanje z vrednostnimi papirji;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UM FNM - Faculty of Natural Sciences and Mathematics
Publisher: [A. Drevenšek]
UDC: 519.22(043.2)
COBISS: 157804035 Link will open in a new window
Views: 50
Downloads: 9
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Mechanical trading system based on reinforcement learning
Secondary abstract: Artificial intelligence and machine learning are becoming part of our everyday life. The situation is similar in financial markets, where so-called intelligent agents trade securities. In this Master's thesis an example of such a mechanical trading system for stock trading is discussed. The model is based on reinforcement learning and uses real, freely available stock market data. In the first part of the thesis, the basic concepts of probability and artificial neural networks are studied. Further, the reinforcement learning and the mathematical background of reinforcement learning are defined in more detail. The second part of the Masterʼs thesis presents the implemented model of the mechanical trading system. Four intelligent agents, which differ in terms of reward systems, are presented and tested. The agents are tested and compared, using a passive strategy.
Secondary keywords: master theses;machine learning;reinforcement learning;artificial neural network;intelligent agent;trading financial instruments;Strojno učenje;Vrednostni papirji;Nevronske mreže (računalništvo);Univerzitetna in visokošolska dela;
Type (COBISS): Master's thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za naravoslovje in matematiko, Oddelek za matematiko in računalništvo
Pages: VIII, 67 f.
ID: 19342310