Strojno učenje za vrednovanje slučajnih zahtjeva

master's thesis

• Title (english): Machine Learning for Contingent Claims Pricing
• Author: Hrvoje Abramović
• Mentor: Zvonko Kostanjčar (mentor)
• Committee member: Zvonko Kostanjčar (predsjednik povjerenstva)
• Committee member: Dražen Jurišić (član povjerenstva)
• Committee member: Stjepan Begušić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-15, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Diplomski rad se bavi primjenom metoda strojnog učenja, točnije LSTM-ova, u svrhu vrednovanja opcija. Ideja je predvidjeti buduću volatilnost i onda provjeriti koje su opcije na tržištu krivo procijenjene gledajući njihovu impliciranu volatilnosti. Model je treniran nad povjesnim podacima S\&P 500 indeksa, te je uspio nadmašiti jednostavan heuristički pristup. Nadalje, model pokazuje najbolje rezultate kada mu se na ulaz, uz povijesnu volatilnost, dovedu i povrati. Pokazalo se i da model radi bolje predikcije nad dužim vremenskim okvirima, što je vjerojatno uzrokovano svojstvom volatilnosti da se dugoročno uvijek vraća na prosječnu vrijednost. Radom su obuhvaćeni i uvodi u opcije, vrednovanje opcija, volatilnost, strategije za trgovanje volatilnošću, te pregled raznih metoda strojnog učenja za vrednovanje opcija. Na kraju je predstavljen i kompletan model za trgovanje volatilnošću pomoću predikcija LSTM modela.

Abstract (english)

This thesis explores the use of machine learning models, namely long-short term memory networks, in the context of option pricing. This is done by trying to forecast future realized volatility, and then determine which options in the market are overpriced or underpriced in terms of their implied volatility. An LSTM model was trained using the historical S\&P 500 data and managed to outperform a simple benchmark model based on historical volatility. It is also shown that the model performs best when returns are passed to the network alongside historical volatility. Furthermore, it is shown that the model performs better with longer forecasting horizon, probably due to the mean reverting property of volatility. Throughout the thesis, introductions to options, option pricing, volatility properties, and volatility trading strategies are presented, as well as an overview of different machine learning approaches used to price options.
In the end, a volatility trading pipeline which employs the LSTM model’s prediction is presented, with a brief guideline for further development.

• Keywords: Opcije; Vrednovanje opcija; Volatilnost; LSTM; Predviđanje volatilinosti; Strategije trgovanja
• Keywords (english): Options; Option pricing; Volatility; LSTM; Volatility forecasting; Trading strategies
• Language: english
• URN:NBN: urn:nbn:hr:168:916331
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 08:17:00