Chapter 1

Introduction
1. 1 Background
One of the most well-known cryptocurrencies had a market value of over $1
trillion in January 2022, with Bitcoin controlling a 67.79% stake [52]. A
CoinDesk case study [53] predicts that the global cryptocurrency market would
expand by 12.9% by 2030. The global cryptocurrency market will expand at a
30% compound annual growth rate (CAGR) from 2019 to 2026 [54]. In one
year, the price of a single Bitcoin increased by 795%, from $7118 in April 2020
to $56,608 in April 2021. Over the past several years, bitcoin values have
unexpectedly increased, making it a viable investment option. To increase
earnings and minimize losses, businesses and investors are focusing more on the
cryptocurrency markets. Purchasing cryptocurrencies for cheap and then selling
them for more money is common practice. Because of this, market experts are
always studying the market to better understand the trends that underlie price
shifts [52, 53, 55]. Financial flows like dividends and earnings, in contrast to
traditional currencies, cannot be used to determine the fundamental value of
cryptocurrencies [56]. Governments and central banks control the financial
system when it comes to traditional currencies. Contrarily, cryptocurrencies
don't have a single controlling body. All transactions are handled and validated
by network nodes using encryption [57]. Then, these will be kept on a
blockchain, a type of open, decentralized ledger. Because of its decentralization,
people remain anonymous throughout transactions. In addition, the supply of
the majority of cryptocurrencies is set and constrained. For instance, there are
only 21 million Bitcoin available, and once that number is reached, no more can
be mined. These variations make it difficult to predict its price swings, which
continue to be a matter of debate [53]. Sentiment significantly affects price
development due to the availability of arbitrage possibilities and intangible
intrinsic worth [52]. Public mood and opinion reflect the erratic nature of
market moves, particularly those involving cryptocurrency [58]. Social media
sentiment is a critical early predictor of future changes in the price of Bitcoin,
according to Yang et al. [59]. This illustrates the potential impact of social
media on one of the biggest market capitalizations. According to a financial
research done by Kristoufek [56], there are other factors impacting bitcoin
values in addition to public sentiment and awareness. Price volatility is
influenced by the hash rate of the cryptocurrency mining process as well as the
relationship between the cryptocurrency and the global financial market. This
thesis is motivated by the notion that, in order to make better investment
decisions, cryptocurrency investors want a reliable price fluctuation prediction
model that makes use of transaction data, sentiment analysis from social media,
blockchain information, and search traffic from search engines. The level of
interest in cryptocurrencies demonstrates how much people believe in Bitcoin
and the technology that underlies it. The majority of the public's attitude has a
significant influence on the future trajectory of the market for cryptocurrency
capitalization as confidence from the public is a factor in the growth of the
market for digital currencies. Therefore, we try to forecast future Bitcoin values
by using public opinion. https://news.google.com A good place to find news
provided by different journalists throughout the world is (Google News).
Additionally, Google News offers the option of searching the news using a set
of keywords, and its search engines additionally offer the capacity to crawl the
news depending on the date of the news release. While Google News presents
the viewpoints of diverse journalists, we also pay attention to popular sentiment.
Reddit, one of the most well-known social media sites where users may post
anonymously, is another option. To forecast the movement of the Bitcoin price,
we also take into account the tone of comments made on Reddit. To understand
the association between all of these characteristics, we have trained many
machine learning models, and the results have been examined.

1 .2 Problem statement
A cryptocurrency is a form of digital money that was first formed as a common
means of exchange. It uses cryptographic techniques like SHA-256 and MD-5
to safeguard the secrecy of financial transactions. Money transfers now require
the involvement of third parties like banks in order to be executed, but bitcoin
does away with this need. Today, our culture accepts the use of
cryptocurrencies. Although several studies have addressed the difficulty of
forecasting changes in stock market prices and the development of successful
trading strategies based on such forecasts, it is crucial to confirm the
applicability of such research in emerging markets, notably the cryptocurrency
market. Strong instability, no interrupted transaction times, relatively little
capitalisation, and high market information availability are the characteristics of
this. The economic viability of the cryptocurrency sector in relation to other
exchanges has already been established, and the mechanisms by which
cryptocurrencies operate have also been established in other industries. Even if
Asian economies have a significant effect on the cryptomarket, it seems to
operate independently of other financial systems. The fact that the technology
used for mining cryptocurrencies offer an alternative to far more established
commodities like gold is one of the market's attractions. These characteristics
have attracted a lot of money, but there hasn't been much study done to create
profitable trading strategies for the bitcoin market. The abundance of publicly
available sentiment data, particularly from social media sites, is a further issue
of discussion in the bitcoin market. This knowledge may conceivably be used to
predict future people's behavior and, as has been shown in past attempts to
predict speculative bubbles in the bitcoin market using emotion polarity, be
used to develop stock trading strategies. Long thought to be a challenging topic,
stock prediction has drawn interest from both academics and traders. Pricing in
the cryptocurrency market contains and reflects chaotic dynamics, in contrast to
returns, and the uncertainty level in returns significantly increased throughout
the period of the increasing regime.The complexity of the work may
undoubtedly be a result of the many variables and uncertainties that influence
the exchanges, such as the economic and political climates as well as
interpersonal relationships. It's challenging, but not impossible, to foresee
changes in stock prices. Because cryptocurrency can be traded on exchanges,
the framework for predicting its price can help investors decide whether to buy,
sell, or hold their cryptocurrency in order to make the most money. It can also
be used by economists and policymakers to study the behavior of the crypto
market. Because the moving components of the cryptocurrency sector are
driven by sentiment through many sources, such as the internet and social
media, it has been shown that sentiment plays a significant influence in
forecasting bitcoin values. As a result, developing techniques that can
accurately identify sentiment indicators from text sources and gauge sentiment
intensity is an essential first step. Studies using deep learning to forecast bitcoin
prices are still in their early stages. Deep learning algorithms have been
employed by several academics to anticipate bitcoin values. Building a precise
methodology for projecting bitcoin values is still crucial, though.
1 .3 Research hypothesis and questions

The paper's research questions are as follows:

 How can deep learning algorithms helps companies and policymakers

estimate cryptocurrency prices?

 What is the right model for estimating future cryptocurrency prices?

 What is the impact of public sentiment on the future of cryptocurrencies

stock price?

 Can deep learning techniques forecast cryptocurrency values accurately?

 What is the best way for validating cryptocurrency price forecasting

model?

1 .4 Aim of studies

This research paper is an attempt to provide powerful indicators for

cryptocurrency exchanges. While cryptocurrency is indeed an intriguing
financial notion, further study of their behavior is required to evaluate their
feasibility as an innovative form of payment. We believe its current value is
influenced by speculation instead of its fundamental worth. The extreme
volatility in the value of cryptocurrency creates confusion for both traders and
those planning to utilize it as money. Because cryptocurrency prices will not act
like conventional banking, it is hard to comprehend what causes this volatility.
As a result, accurately predicting the future price of any cryptocurrency is
difficult. Nevertheless, because it is such a high-risk investment, risk
management is critical. As a result, forecasting the fluctuation of the crypto
market, which would be linked to high market risks, is required to assist
investors in managing their investments or effectively preparing for future risks.
Correct predictions can help cryptocurrency investors make better investment
choices and potentially maximize profits. They can indeed help policymakers
and economic investigators investigate the behavioral patterns of
cryptocurrency markets. Nonetheless, cryptocurrency cost forecasting is
regarded as a hard challenge due to its chaos and complicated nature. People's
belief in cryptocurrency and the underlying technology is directly proportional
to their desire for cryptocurrencies. Because the public's faith is engaged in the
growth of the cryptocurrency industry, public sentiment has a significant effect
on the future of cryptocurrency stock prices. So people engage in the markets
based on general feelings and technical analyses. As a result, taking human
feelings into account can provide a high level of prediction. The aim of this
study is to develop an accurate and reliable bitcoin price prediction model using
deep learning and sentiment analysis techniques. Specifically, the thesis will
analyze data collected from various sources, such as Reddit and Google News,
to identify the correlation between cryptocurrency sentiment and price
movement. The primary objective is to offer insights into the potential impacts
of market sentiment on cryptocurrency price volatility. The study will be
valuable to investors, traders, and analysts seeking to make informed decisions
in the cryptocurrency market. Ultimately, the goal is to provide a
comprehensive, accessible, and user-friendly resource for understanding the role
of sentiment analysis in bitcoin price prediction modeling.

1. 5 practical use of the research

Cryptocurrency price prediction could indeed assist cryptocurrency stakeholders
in making better financial decisions in order to maximize profits. Currently, the
cryptocurrency market is by far the most ideal venue for speculative trading.
Many people have made a significant amount of money by trading in digital
marketplaces and also investing in advertising. Cryptocurrency pricing
movement analysis is critical for investors because it advises them on the best
moment to start trading. It also assists investors in determining whether to
purchase, sell, or keep cryptocurrencies to maximize profits. Predicting
cryptocurrency prices may aid in understanding and forecasting probable
market fluctuations and humankind's future progress.
1 .6 Innovation aspect of research

1 . 7 Methodology of research

1 . 8 Research method type

The research method type for predicting bitcoin price using deep learning
models and sentiment analysis is quantitative research. This is because the study
involves the use of mathematical models and statistical techniques to analyze
data and make predictions.
1 . 9 Structure of Thesis
This thesis is organized as follows:

Chapter 2: we explore the subject of cryptocurrencies, provide a technical

overview of them, and delve into the market dynamics.

Chapter3 : This chapter explores one of the most popular machine learning
algorithms, neural networks, and describes how they operate mathematically.
followed by a primer on deep learning algorithms.

Chapter 4 : This chapter explains how to put this thesis into practice, how we
collected and organized our data, what approach we made for the model for
prediction, and how we predicted prices.

Chapter5 : In this chapter, we assess the accuracy of our prediction model's

predictions. It brings our research to a close while taking a broad look at the
results.