Incremental reinforcement learning for portfolio optimisation.

Simple item page
dc.contributor.advisorAndries, Engelbrechten_ZA
dc.contributor.authorRefiloe, Shabeen_ZA
dc.contributor.otherStellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering.en_ZA
dc.date.accessioned2023-01-30T19:26:30Zen_ZA
dc.date.accessioned2023-05-18T07:00:01Zen_ZA
dc.date.available2023-01-30T19:26:30Zen_ZA
dc.date.available2023-05-18T07:00:01Zen_ZA
dc.date.issued2023-03en_ZA
dc.descriptionThesis (MEng)--Stellenbosch University, 2023.en_ZA
dc.description.abstractENGLISH ABSTRACT: Portfolio optimisation is a decision-making problem that involves allocation of a certain fund across different financial assets, with the objective of maximising profit and minimising risk, simultaneously. Portfolio optimisation is a difficult problem to analyse. There is a wide range of research on various portfolio optimisation approaches in finance and computational intelligence. The two fields overlap. Thus, the use of meta-heuristics to make intelligent investment decisions is a result of the intersection of finance and computational intelligence. Meta-heuristics formulated the portfolio optimisation problem as a static optimisation problem and successfully obtained optimal portfolios. However, in the real world, investment decision-making is a dynamic problem that involves daily trading. Therefore, it is more representative of real-world investments to formulate the portfolio optimisation problem as a dynamic optimisation problem. This thesis explores a reinforcement learning approach to formulate a dynamic investment strategy. The concept of reinforcement learning has improved the development of multistage stochastic optimisation; a primary component in sequential portfolio optimisation. A recurrent form of a reinforcement learning algorithm called proximal policy optimisation (PPO), that allocates portfolios based on historic asset prices is presented. The results provide a conclusive support for the ability of PPO to identify good-quality portfolios. The results also show that the strategy becomes outdated overtime as it fails to perform as well during the COVID-19 pandemic. Based on this finding, the recurrent PPO approach was improved in order to take into account the presence of concept drift caused by pandemics and potential financial contagions. The approach was adapted to incrementally learn the financial market as the portfolio optimisation process takes place. The incremental recurrent PPO algorithm is shown to be able to adapt to drastic changes in the market and obtain optimal portfolios. en_ZA
dc.description.abstractAFRIKAANS OPSOMMING: Portefeulje-optimering is ’n besluitnemingsprobleem wat die toewysing van ’n sekere fonds in verskillende finansi¨ele bates behels, terwyl die wins terselfdertyd maksimeer en risiko geminimaliseer word. Portefeulje-optimering is ’n moeilike probleem om te ontleed. Daar is ’n wye reeks navorsing oor verskeie portefeulje-optimaliseringsbenaderings in finansies en rekenaarintelligensie. Die twee velde oorvleuel. Dus, die gebruik van metaheuristiek om intelligente beleggingsbesluite te neem is ’n gevolg van die kruising van finansies en rekenaarintelligensie. Metaheuristiek het die portefeulje-optimeringsprobleem as ’n statiese optimeringsprobleem geformuleer en optimale portefeuljes suksesvol verkry. In die regte wˆereld is beleggingsbesluitneming egter ’n dinamiese probleem wat daaglikse handel behels. Daarom is dit meer verteenwoordigend van werklike beleggings om die portefeuljeoptimeringsprobleem as ’n dinamiese optimaliseringsprobleem te formuleer. Hierdie tesis ondersoek ’n versterkingsleerbenadering om ’n dinamiese beleggingstrategie te formuleer. Die konsep van versterkingsleer het die ontwikkeling van meerfase stogastiese optimering verbeter; ’n primˆere komponent in opeenvolgende portefeulje-optimering. ’n Herhalende vorm van ’n versterkende leeralgoritme genaamd proksimale beleidsoptimering (PPO), wat portefeuljes toewys op grond van historiese batepryse, word aangebied. Die resultate bied ’n afdoende ondersteuning vir die vermo¨e van herhalende PPO om goeie gehalte portefeuljes te identifiseer. Die resultate toon ook dat die strategie oortyd verouderd raak aangesien dit nie so goed presteer tydens die COVID-19-toediening nie. Op grond van hierdie bevinding is die herhalende PPO-benadering verbeter om die teenwoordigheid van konsepverskuiwing wat deur pandemies en potensi¨ele finansi¨ele besmettings veroorsaak word, in ag te neem. Die benadering is aangepas om die finansi¨ele mark inkrementeel te leer namate die portefeulje-optimeringsproses plaasvind. Daar word getoon dat die inkrementele herhalende PPO-algoritme in staat is om aan te pas by drastiese veranderinge in die mark en optimale portefeuljes te verkry.af_ZA
dc.description.versionMastersen_ZA
dc.format.extentxi, 111 pages : illustrations.en_ZA
dc.identifier.urihttp://hdl.handle.net/10019.1/127011en_ZA
dc.language.isoen_ZAen_ZA
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : Stellenbosch Universityen_ZA
dc.rights.holderStellenbosch Universityen_ZA
dc.subject.lcshPortfolio management -- Mathematical modelsen_ZA
dc.subject.lcshReinforcement learningen_ZA
dc.subject.lcshMathematical optimizationen_ZA
dc.titleIncremental reinforcement learning for portfolio optimisation.en_ZA
dc.typeThesisen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
shabe_incremental_2023.pdf
Size:
2.42 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Masters Degrees (Industrial Engineering)