izpis_h1_title_alt

Uporaba modelov ARIMA in GARCH za modeliranje tržišča kriptovalut in razvoj trgovalne strategije
ID Balantič, Domen (Author), ID Lavbič, Dejan (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (2,08 MB)
MD5: 37D361F1CFA28338141C24F0CC07C708

Abstract
V magistrskem delu smo se spoznali z analizo in modeliranjem podatkov, predstavljenih s pomočjo časovnih vrst. Pri tem smo spoznali in uporabili različne matematične metode ter se poglobili v analizo modelov ARIMA in GARCH. Pridobljeno znanje smo uporabili za razvoj osnovne trgovalne strategije v programskem jeziku R. Uporaba kombinacije modelov ARIMA in GARCH se je na klasičnih finančnih inštrumentih izkazala kot uspešna, primernost uporabe na tržiščih kriptovalut pa še ni bila širše raziskana. Zato smo v magistrskem delu razvito trgovalno strategijo nadalje aplicirali na gibanju tržišča kriptovalut, natančneje na gibanju cene kriptovalute Bitcoin. Delovanje osnovne strategije smo poskušali izboljšati z različnimi pristopi, med drugim tudi z indikatorji, specifičnimi za delovanje kriptovalute Bitcoin. Med izboljšavami smo skušali upoštevati delovanje in obnašanje klasičnih finančnih trgov. Analizirali in opisali smo doprinos posameznih pristopov ter z njimi skušali izboljšati delovanje osnovne trgovalne strategije. Za lažjo oceno delovanja smo definirali referenčne strategije in različne metrike, ki smo jih nato uporabili za primerjavo donosov. Nazadnje smo, glede na rezultate primerjave z referenčnimi strategijami in rezultate metrik za ocenjevanje uspešnosti strategij, določili najuspešnejšo strategijo, ki bi jo lahko uporabili v sistemu za samodejno trgovanje s kriptovalutami. Glede na rezultate metrike čistega dobička smo ugotovili, da razvita strategija AGLN dosega zelo visoko donosnost - 4,55-krat višjo kot strategija KUPI in ZADRŽI in 1,5-krat višjo donosnost kot strategija križanja MACD. Od tod lahko sklepamo, da je uporaba kombinacije modelov ARIMA in GARCH primerna tudi za modeliranje tržišča kriptovalut, z upoštevanjem njegovih lastnosti in posebnosti.

Language:Slovenian
Keywords:ARIMA, GARCH, Bitcoin, trgovalna strategija, napovedovanje gibanja
Work type:Master's thesis/paper
Typology:2.09 - Master's Thesis
Organization:FMF - Faculty of Mathematics and Physics
FRI - Faculty of Computer and Information Science
Year:2020
PID:20.500.12556/RUL-116869 This link opens in a new window
COBISS.SI-ID:19721475 This link opens in a new window
Publication date in RUL:14.06.2020
Views:1874
Downloads:330
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Application of ARIMA and GARCH models for cryptocurrency market modeling and trade strategy development
Abstract:
In this master's thesis we have learned about time series analysis and modelling. We have examined different mathematical methods and focused on ARIMA and GARCH model analysis. Based on gained knowledge we have developed basic trading strategy in R programming language. Combination of ARIMA and GARCH models has already been proven successful on classical financial markets but wasn't yet widely explored for cryptocurrencies. Therefore, the developed trading strategy was applied on Bitcoin cryptocurrency market data. With behaviour of classical financial markets in mind, we have tried to improve our basic strategy with different approaches, including indicators specific to behaviour of Bitcoin cryptocurrency. We have analysed and described contribution of individual approaches and tried to improve developed strategy with them. To help evaluate performance, we defined reference strategies and different metrics, which were then used to compare returns. Based on benchmarking and metrics results, we have identified the most successful strategy, that could be further used in an automated trading system. We have discovered that the AGLN trading strategy achieves very high profitability - 4,55 times higher than BUY and HOLD strategy and 1,55 times higher than MACD crossover strategy. Furthermore, we can conclude that the combination of ARIMA and GARCH models is also suitable for modelling cryptocurrency markets.

Keywords:ARIMA, GARCH, Bitcoin, trading strategy, forecasting

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back