Razvoj inteligentnih agentov za trgovanje s kriptovalutami

LEBAN, JAKA (author), Štruc, Vitomir (mentor)

Kripto trgi so izjemno volatilni, cene kripto žetonov nihajo neprimerno bolj kot pri borznih trgih. Zato je smiselno razmisliti o avtomatskem trgovanju kriptovalut. Cilj tega projekta je razviti inteligentnega agenta, ki preko API povezav v realnem času komunicira s kripto borzami in na podlagi analiz izvršuje nakupe in prodaje kriptovalut. Ker je cena kriptovalute določena na podlagi ponudbe in povpraševanja, je največja težava ovrednotiti njeno realno vrednost. V primeru delnic podjetij se lahko poslužujemo temeljne analize in ocenimo vrednost podjetja, produktov in delovne sile ter tako ocenimo njihovo vrednost. Pri kriptovalutah ta analiza največkrat odpove ali pa lahko z njo zgolj ocenimo potencial produkta. V tej magistrski nalogi zato raziskujemo možnost napovedovanja cen kritptovalut s tehnično in Twitter analizo. V okviru dela izdelamo inteligentnega agenta, s katerim lahko uporabnik spremlja ceno izbrane kriptovalute, različne tehnične analize ali napovedi strojnega učenja. Poleg omenjenih analiz inteligentni agent podpira Twitter analizo razpoloženja tvitov vplivne osebe in splošno analizo razpoloženja. Inteligentni agent podpira nekaj enostavnejših algoritmov trgovanja, ki na podlagi omenjenih analiz izvršujejo nakupe in prodaje kripto žetonov. Algoritmi trgovanja na podlagi tehnične analize izhajajo iz predlogov ekonomistov in strojnega učenja, algoritmi trgovanja pri Twitter analizi pa bazirajo na razpoloženju tvitov vplivne osebe. Rezultati strategij so analizirani in predstavljeni na nekaterih izbranih simulacijah trgovanja.

• Language: Slovenian
• Keywords: kriptovalute, inteligentni agenti, tehnična analiza, regresija, strojno učenje, Twitter analiza, razpoloženje tvitov
• Work type: Master's thesis/paper
• Organization: FE - Faculty of Electrical Engineering
• Year: 2020
• PID: 20.500.12556/RUL-115961
• Publication date in RUL: 03.05.2020

Secondary language

• Language: English
• Title: Development of intelligent agents for cryptocurrency trading

Abstract

Crypto markets are extremely volatile and cryptocurrency prices tend to fluctuate much more than stock prices. Therefore, automatic algorithm trading is a sensible solution. The goal of this project is to build an intelligent agent that enables constant communication with the crypto exchanges and uses its analyses to execute trades. Since supply and demand usually dictate the price of any cryptocurrency, the biggest problem is how to determine its real value. In the case of company shares, fundamental analysis can be used to determine the value of the company, its products and its workforce. Based on that, the stock value of the company shares can be determined. When it comes to cryptocurrencies, fundamental analysis often fails or can only evaluate the potential of a product. For this reason, this Master's degree aims to explore the possibility of predicting cryptocurrencies’ price movements by means of a technical and Twitter analysis. Our main goal is to build an intelligent agent where users can follow the chosen cryptocurrency prices and also check various technical analyses or machine learning forecasts. The intelligent agent also supports both the general Twitter sentimental analysis as well as the sentimental analysis of a specific influencer. The intelligent agent supports several simple trading algorithms, which execute trades in crypto tokens based on the above mentioned analyses. The technical-analysis-based trading algorithms are the result of the economists’ suggestions, whereas the Twitter algorithms are based on the tweet sentiment of a specific influencer. The results of the strategies have been analysed and presented in several chosen trading simulations.

• Keywords: cryptocurrencies, intelligent agents, technical analysis, regression, machine learning, Twitter analysis, tweets sentimental