Predikce vývoje ceny vybraných kryptoměn pomocí strojového učení

Abstract

Radu let je oblasti velkeho zajmu jak akademiku, tak ekonomu predikce financnich casovych rad. Potencial spravne predikovat aktiva libovolneho druhu se dramaticky zlepsil s nastupem umele inteligence. Specificky, aplikace strojoveho uceni na casove rady je tematem mnoha diskuzi. Ackoliv bylo vydano nekolik vedeckych praci a clanku na zminene tema, pouze hrstka se zabyva uzitim umele inteligance v predikci cen jednotlivych kryptomen. Tato bakalarska prace se zabyva aplikaci algoritmu strojoveho uceni na klicove technicke indikatory a jejich ucinnosti v predikci vybrane kryptomeny. Modely a indikatory jsou naprogrogramovane v Pythonu. V zaveru teto prace se porovnaji vysledky jed- notlivych predikcnich modelu.

The aim of this bachelor's thesis is to investigate the efficiency and application of machine learning algorithms to predict the future price of a cryptocurrency. The first half of the paper deals with a brief description of cryptocurrencies and the technology backing the whole ecosystem. Following an evaluation of traditional methodologies used to forecast financial times series and how the cryptocurrency market reacts to the application of these methods. Appropriate features are engineered from data-sets of selected cryptocurrencies and applied to the machine learning algorithms. These algorithms are eventually tested. Their performances are consequentially compared and a conclusion is drawn from the results.