20.500.12556/RUL-138779 Napoved gibanja delniških tečajev in kriptovalut z uporabo javnega sentimenta na socialnem omrežju Reddit Prediction of stock and cryptocurrencies movements using public sentiment from Reddit posts Diplomsko delo poskusa odgovoriti na vprasanje, ce in kako hobijske skupine na socialnem omrezju Reddit vplivajo na rast in padanje cen delnic podjetij in kriptovalut, katerih zagovorniki oz. kritiki so. Analizirali smo objave iz tistih skupin omrezja Reddit, ki se ukvarjajo z diskusijo in analizo relevantnih delnic in kriptovalut. Financne podatke in cene trznih nancnih sredstev, katerih padec ali rast modeliramo, smo pridobili na spletnem viru Yahoo Finance [18]. Za analizo sentimenta na socialnem omrezju smo uporabili python knjiznici TextBlob in VADER [14][17], podobno kot v drugih diplomskih delih [43]. Opravili smo analizo vzrocnosti, da preverimo katere znacilnosti najbolje napovejo koncno ceno opazovane delnice oz. kriptovalute. Na podlagi dobljenih znacilnosti in premikanja cen delnic ter kriptovalut, smo sestavili skupino napovednih modelov, in jih po ustreznosti primerjali med sabo. Koncni model naj bi bil za investitorje uporaben vzorec, ki z neko natancnostjo predvidi gibanje neke delnice v bliznji prihodnosti, glede na njeno gibanje in sentiment v bliznji preteklosti. Analiza nakazuje, da navdusenje nad doloceno delnico oz. kriptovaluto, merjeno s povecanim stevilom objav v tistem casu, vsebuje pomembne informacije glede gibanja cene tega trznega sredstva, ce opazovani komentatorji predstavljajo dovolj velik delez investitorjev tega trznega sredstva. Ocena sentimenta ne kaze nobenega vpliva. The thesis attempts to answer the question, if - and to what extent - online in uencers on the social media site Reddit in uence the movements of stocks and cryptocurrencies, whose companies they advocate or criticise. We have analyzed the submissions of those subreddits, that deal with discussion and analysis of relevant stocks and cryptocurrencis. The stock market data for the chosen assets was found on the Yahoo Finance website [18]. For sentiment analysis of submissions, we used python libraries TextBlob and VADER [14][17], as they have been used eectively in other works [43]. Using Granger Causality analysis, we selected the best-performing features. On the basis of given values and the movement of stocks, we have built a group of predictive models, which we validated and compared to each other to get the best result. The nal model would be a useful tool for investors, which predicts patterns of stock market movement in the near future, based on its movement and sentiment in the near past. Our analysis suggests, that enthusiasm about a particular asset - measured with an increase in the number of posts about it - contains important information about the movement of said asset, if the post authors represent a large enough share of the investors of that asset. kriptovaluta socialna omrežja borza delnice analiza sentimenta strojno učenje Reddit cryptocurrency social media stock market stocks sentiment analysis machine learning Reddit true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2022-08-17 16:10:01 2022-08-17 16:10:20 2022-09-15 04:17:09 0000-00-00 00:00:00 2022 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 35255 120422403 Selic_Jon_-_Napoved_gibanja_delniskih_tecajev_in_kriptovalut_z_uporabo_javnega_sentimenta_na_soc.pdf Selic_Jon_-_Napoved_gibanja_delniskih_tecajev_in_kriptovalut_z_uporabo_javnega_sentimenta_na_soc.pdf 1 23B6271392C69E3077313A58ADEF7B71 c95f7023c6812859fa7b31b6949bbdbd542e2571442b58498557dba7f540e95b 5274ac24-1e36-11ed-92af-00155dcfd717 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=159229 Fakulteta za računalništvo in informatiko 0 0 0