Article

A Novel Cryptocurrency Price Prediction Model Using GRU,

LSTM and bi-LSTM Machine Learning Algorithms
Mohammad J. Hamayel and Amani Yousef Owda *

Department of Natural, Engineering and Technology Sciences, Arab American University,

Ramallah P600, Palestine; [email protected]
* Correspondence: [email protected]

Abstract: Cryptocurrency is a new sort of asset that has emerged as a result of the advancement
of financial technology and it has created a big opportunity for researches. Cryptocurrency price
forecasting is difficult due to price volatility and dynamism. Around the world, there are hundreds
of cryptocurrencies that are used. This paper proposes three types of recurrent neural network
(RNN) algorithms used to predict the prices of three types of cryptocurrencies, namely Bitcoin
(BTC), Litecoin (LTC), and Ethereum (ETH). The models show excellent predictions depending on
the mean absolute percentage error (MAPE). Results obtained from these models show that the
gated recurrent unit (GRU) performed better in prediction for all types of cryptocurrency than the
long short-term memory (LSTM) and bidirectional LSTM (bi-LSTM) models. Therefore, it can be
considered the best algorithm. GRU presents the most accurate prediction for LTC with MAPE
percentages of 0.2454%, 0.8267%, and 0.2116% for BTC, ETH, and LTC, respectively. The bi-LSTM
algorithm presents the lowest prediction result compared with the other two algorithms as the
MAPE percentages are: 5.990%, 6.85%, and 2.332% for BTC, ETH, and LTC, respectively. Overall, the



prediction models in this paper represent accurate results close to the actual prices of cryptocurrencies.

 The importance of having these models is that they can have significant economic ramifications
Citation: Hamayel, M.J.; Owda, A.Y. by helping investors and traders to pinpoint cryptocurrency sales and purchasing. As a plan for
A Novel Cryptocurrency Price future work, a recommendation is made to investigate other factors that might affect the prices of
Prediction Model Using GRU, LSTM cryptocurrency market such as social media, tweets, and trading volume.
and bi-LSTM Machine Learning
Algorithms. AI 2021, 2, 477–496. Keywords: cryptography; blockchain; cryptocurrency; artificial intelligence (AI); machine learning
https://doi.org/10.3390/ai2040030

Academic Editor: Amir Mosavi

1. Introduction
Received: 8 September 2021
Accepted: 5 October 2021
Cryptocurrency is a virtual or digital currency used in financial systems [1,2]. It is
Published: 13 October 2021
secured by cryptography that makes it impossible to be counterfeited or double-spent.
Furthermore, it is not issued from a central authority or central banks, and it is decentralized
Publisher’s Note: MDPI stays neutral
virtual currencies that can be converted via cryptographic procedures [3] and this make
with regard to jurisdictional claims in
it distinguishable from traditional currencies. The other feature is that it is created by
published maps and institutional affil- technology called blockchain [4], which is an extremely complex, and aims to storing data
iations. that makes it difficult or impossible to alter, hack, or defraud the system. Bitcoin has begun
to carve out a niche for itself, which may either help cryptocurrencies to gain widespread
acceptance or be the major cause of their demise. Cryptocurrencies are still in their infancy,
and it is difficult to predict whether they will ever be widely used in global markets or
Copyright: © 2021 by the authors.
not [5]. The most prominent cryptocurrency, Bitcoin, was established in 2009 and for more
Licensee MDPI, Basel, Switzerland.
than two years was the sole Blockchain-based cryptocurrency. Today, however, there are
This article is an open access article
over 5000 cryptocurrencies and 5.8 million active users in the cryptocurrency industry [6].
distributed under the terms and Because of its intrinsic nature of mixing encryption technology with monetary units, Bitcoin
conditions of the Creative Commons has recently gotten a lot of attention in the disciplines of economics, cryptography, and
Attribution (CC BY) license (https:// computer science [7].
creativecommons.org/licenses/by/ Blockchain (BC), the technology that underpins the Bitcoin cryptocurrency system, is
4.0/). widely seen as critical in providing the backbone for assuring greater security and privacy

AI 2021, 2, 477–496. https://doi.org/10.3390/ai2040030 https://www.mdpi.com/journal/ai

AI 2021, 2 478

in a variety of other fields, including the Internet of Things (IoT) ecosystem. It is mainly
a digital ledger of transactions that is distributed across the entire network of computer
systems on the blockchain [8,9]. The blockchain consists of two fundamental components;
the first one is a transaction, and the second is a block. The transaction represents the action
triggered by the participant, and the block is a data collection that records the transaction
and additional details such as the correct sequence and creation timestamp. Blockchain
have a signaling system (BloSS) of multi-domain, blockchain-based, cooperative DDoS
defense system in which each autonomous system (AS) joins the defensive alliance [10].
Reference [11] reveals that the effects of networks on competition in the nascent cryp-
tocurrency market over a period of time regarding exchange rates among cryptocurrencies
depends on two aspects: (1) competition among different currencies and (2) competition
among exchanges. There are hundreds of cryptocurrencies, but Bitcoin is the most popular
one as it is a stubborn competitor and did not emerge out of the cryptocurrency competition
track. As a result, it has become the dominant cryptocurrency. The authors of [12] describe
the competition between cryptocurrency as “healthy competition” and suggests that new
technology and security innovation. The authors of [13] reveal that Bitcoin and national
currencies show volatility shock transmission, while economic policy uncertainty has little
effect. The authors of [14] investigate the interaction between big data and cryptocurrency.
One of the most appealing marketplaces for financial speculation is the cryptocurrency
market, which means that deceptive activities have flourished via social media. Many
people have reaped a lot of profits through speculation in the digital markets, but every
investment process suffers from many hidden risks and some investors, particularly those
with a high risk tolerance, are interested in investing in cryptocurrency. Therefore, market
analysts and speculators rely on prediction [15,16]. With variations in predictive power
per cryptocurrency, machine learning and artificial intelligence algorithms are moderately
appealing [17]. Low-volatility cryptocurrencies are more predictable than high-volatility
ones. There is evidence that the usefulness of different information sets varies between ma-
chine learning algorithms, implying that prediction is likely to be much more complicated
when a set of machine learning algorithms is used [18]. Despite the widespread use of
cryptocurrencies for various types of purchases and transactions around the world, there is
no consistent opinion on the definition of cryptocurrency or its legal status [19]. Further-
more, the aforementioned situation exacerbates challenges in criminal investigations of
cryptocurrency-based money laundering. As a result, law enforcement organizations are
having difficulty pinpointing criminals’ identities and proving that they have committed a
crime [20]. Focusing on Bitcoin pricing is similar to stock pricing: none of the risk variables
that explain stock price movements apply to cryptocurrencies. Furthermore, traditional
macroeconomic variables such as currency rates, commodity prices, and macroeconomic
factors that affect other assets have little to no impact on most cryptocurrencies [21]. As a
result of the cryptocurrency market’s surge in 2017 [22], various governments across the
world have begun to move toward standardizing and overseeing digital money. Because
of the security of blockchain technology and their economic environment, people have
become more confident in using Bitcoin [22]. Although the blockchain provides a high
security ecosystem, the research area surrounding the legality of cryptocurrency cannot
be isolated from the people who utilize cryptocurrencies for illicit purposes. The legality
of cryptocurrencies has been the subject of numerous debates. The authors of [23] dis-
cuss the perspectives and the nature of cryptocurrencies in terms of monetary features,
legal considerations, economic considerations, and Sharia considerations. Based on the
perspective and characteristics of traditional currency, cryptocurrency does not satisfy the
characteristics of a currency from an economic standpoint.
There are hundreds of cryptocurrencies in digital markets, but Bitcoin is the most
popular and is affected and interacted with by external influences such as the news, social
media, and small cryptocurrencies that have a limited market share, which are often not
taked into account from investors and traders. Due to the strong relationships between
cryptocurrencies, the smaller ones have become a source of shocks that can positively
AI 2021, 2 479

or negatively affect other cryptocurrencies. The authors of [24] reveal that gold as an
independent currency can be used as a good hedging instrument to decrease the risk
related to unexpected movement in the cryptocurrency market.
Cryptocurrency prices are difficult to forecast due to price volatility and dynamism.
Around the world there are hundreds of cryptocurrencies that clients use. In this paper,
we focus on three of the most popular ones. As a result, the paper aims to achieve the
following by using deep leaning algorithms, which can discover hidden patterns from data,
integrate them, and create far more efficient predictions:
• Presenting a comprehensive study of the various existing schemes to predict the prices
of BTC, ETH, and LTC cryptocurrencies.
• Using AI algorithms such as LSTM, bi-LSTM, and GRU to accurately predict the prices
of cryptocurrencies.
• Utilizing long short-term memory (LSTM), a deep learning algorithm, and Fbprophet,
which is an auto machine learning algorithm, for prediction.
• Evaluating the proposed hybrid models using evaluation matrices such as RMSE and
MAPE for Bitcoin, Ethereum, and Litecoin.
The main idea behind these models is to achieve a reliable prediction model that
investors can rely on based on historical cryptocurrency prices. Moreover, the paper aims
to answer the following research questions: ‘How can machine learning algorithms help
investors and decision makers to predict cryptocurrency prices?’ and ‘What is the best
model for predicting future cryptocurrency prices?’
This section provides an overview of cryptocurrencies and the remainder of the paper
is structured as follows: Section 2 describes the literature review and previous work in
this field, Section 3 presents the modeling results and the statistical analysis of the data,
Section 4 describes the used dataset, Section 5 illustrates the experimental results, Section 6
presents a comparison between the model proposed in this paper and those of other studies
in the literature, and Section 7 summarizes the overall conclusions of the paper.

2. Literature Review
Machine learning (ML) is a type of artificial intelligence that can predict the future
based on past data. ML-based models have various advantages over other forecasting
models as prior research has shown that it not only delivers a result that is nearly or exactly
the same as the actual result, but it also improves the accuracy of the result [25]. Examples
of machine learning include neural networks (NN), support vector machines (SVM), and
deep learning. The authors of [26] demonstrate that incorporating cryptocurrency into
a portfolio improves its effectiveness in two ways. The first is to reduce the standard
deviation, and the second is to provide investors with more allocation options. The best
cryptocurrency allocation was reported to be in the range from 5% to 20%, depending on
the risk tolerance of the investor. The authors of [27] focus on time series data forecasting in
particular and apply two machine learning algorithms, random forests (RF) and stochastic
gradient boosting machine (SGBM). The results show that the ML ensemble technique can
be used to anticipate Bitcoin values.
The decision-making process needs to make the appropriate decision at the right time,
reducing the risks associated with the investment process. In [28], a hybrid cryptocurrency
prediction system based on LSTM and GRU is presented, focusing on two cryptocurrencies,
Litecoin and Monero. The authors of [29] use minute-sampled Bitcoin returns over 3 h
periods to aggregate RV data. A variety of machine learning methods, including ANN
(MLP, GRU, and LSTM), SVM, and ridge regression, were used to predict future values
based on past samples, which are compared to the heterogeneous auto-regressive realized
volatility (HARRV) model with optimized lag parameters. The findings show that the
suggested system correctly predicts prices with high accuracy, indicating that the method
may be used to forecast prices for a variety of cryptocurrencies. The authors of [30] employ
the traditional support vector machine and linear regression methods to forecast Bitcoin
values. This research takes into account a time series prediction made up of everyday
AI 2021, 2 480

Bitcoin closing prices for the creation of Bitcoin prediction models. The authors of [31]
used machine learning techniques to address both a multiple regression technique that
relies on highly correlated characteristics and a deep learning mechanism that uses a
conjugate gradient mechanism in conjunction with a linear search for BTC price prediction.
In [32], the price movements of Bitcoin, Ethereum, and Ripple are analyzed. The authors
utilize powerful artificial intelligence frameworks, including a fully linked artificial neural
network (ANN) and a long short-term memory (LSTM) recurrent neural network, and they
discovered that ANN relies more on long-term history, whereas LSTM relies more on short-
term dynamics, implying that LSTM is more efficient at extracting meaningful information
from historical memory than ANN. The study in [33] on Bitcoin daily price prediction
with high-dimensional data reveals that logistic regression and linear discriminant analysis
achieve an accuracy of 66%. On the other hand, surpassing (a sophisticated machine
learning algorithm) outperforms the benchmark results for daily price prediction, with
statistical techniques and machine learning algorithms having the greatest accuracies of
66% and 65.3%, respectively. The study in [34] examines the use of neural networks (NN),
support vector machines (SVM), and random forest (RF). The findings demonstrate that
machine learning and sentiment analysis may be used to anticipate cryptocurrency markets
(with Twitter data alone being able to predict specific coins) and that NN outperforms the
other models. In [35], the LSTM model is used to predict and find methods for forecasting
Bitcoin on the stock market through Yahoo Finance that may predict a result of more than
12,600 USD in the days after the prediction. Due to the importance of the development of a
robust and reliable method for predicting cryptocurrency prices, researchers have focused
on more innovative models. In [36], both linear and non-linear time-series components of
the stock dataset were used for forecasting using the hybrid model. In the non-linear time
series forecast, CNN and Seq2Seq LSTMs were successfully coupled for dynamic modeling
of short- and long-term dependent patterns. The study in [37] focused on social factors,
which are increasingly being utilized for online transactions throughout the world, by using
a multi-linear regression model and that analyzes two big capital market cryptocurrencies,
BTC and LTC. The authors of [37] found that the R2 scores were 44% for LTC and 59% for
BTC. Ref. [38] used two different LSTM models (a standard LSTM model and an LSTM
with an AR(2) model). This study presented a forecasting framework, using an LSTM
model to forecast Bitcoin daily prices. The study in [38] found that the model with AR(2)
was better than LSTM with an RMSE of 247.33. Researchers in [39] compared three different
models (ARIMA, LSTM and GRU) for predicting BTC’s price. The experimental outcomes
in [39] showed that ARIMA achieved the best performance with a MAPE of 2.76% and
RMSE of 302.53. The study in [40] presented two types of prediction models constructed
using Bayesian optimized RNN and LSTM to predict the price of BTC. The study revealed
that LSTM showed better performance and achieved an accuracy of 52% and RMSE of 8%.
The investment process mainly depends on the historical price of a cryptocurrency.
One of the most important strategies that the investor depends on is building Markov
chains. This strategy consists of multiple decision trees that are used to identify the
cryptocurrency that is estimated to provide a greater return when sold and then comparing
the estimation with the actual figure [41]. Due to the importance of prediction in the
investment process that many people depend on to earn revenue, this paper focuses
on three models that can predict future cryptocurrency prices using machine learning
algorithms and artificial intelligence approaches to achieve accurate prediction models
with the aim of helping investors.

3. Materials and Methods

To achieve the aims of this paper, we trained three distinct models for three different
forms of cryptocurrency price prediction using historical cryptocurrency prices. Then, in
order to evaluate the suggested schemes’ performances, we compare the accuracy of our
proposed model to that of current models by following five stages: (1) collecting historical
 3. Materials and Methods
To achieve the aims of this paper, we trained three distinct models for three different
AI 2021, 2 forms of cryptocurrency price prediction using historical cryptocurrency prices. Then, in 481
order to evaluate the suggested schemes’ performances, we compare the accuracy of our
proposed model to that of current models by following five stages: (1) collecting historical
cryptocurrency data; (2) data exploration and visualization; (3) training three types of
cryptocurrency data; (2) data exploration and visualization; (3) training three types of
models; (4) testing the models; and (5) extracting and comparing the results.
models; (4) testing the models; and (5) extracting and comparing the results.
In this section, we present and compare three types of algorithms—long short-term
In this section, we present and compare three types of algorithms—long short-term
memory (LSTM), gated recurrent unit (GRU), and bidirectional LSTM (bi-LSTM)—to pre-
memory (LSTM), gated recurrent unit (GRU), and bidirectional LSTM (bi-LSTM)—to pre-
dict the price of three types of cryptocurrency based on historical data—Bitcoin (BTC),
dict the price of three types of cryptocurrency based on historical data—Bitcoin (BTC),
Litecoin (LTC) and Ethereum (ETH). Figure 1 shows the methodology of processing the
Litecoin (LTC) and Ethereum (ETH). Figure 1 shows the methodology of processing the
dataset. It starts with data collection, then the data visualization process is used to illus-
dataset. It starts with data collection, then the data visualization process is used to illus-
trate and explore the data’s behavior and distribution and the relationship between the
trate and explore the data’s behavior and distribution and the relationship between the
cryptocurrencies. Next, the Next,
cryptocurrencies. models are
the trained
models with
are 80% with
trained of the80%
collected dataset [42].
of the collected The [42]. The
dataset
training dataset is from 22 January 2018 until 22 October 2020 and the testing dataset (20%
training dataset is from 22 January 2018 until 22 October 2020 and the testing dataset (20%
of the data)
of is form
the 22 is
data) October 2020
form 22 until 30
October June
2020 2021.
until 30Then, after training
June 2021. the models
Then, after trainingwethe models
tested them.
we tested them. Then, we extracted and compared the results and selectedmodel
Then, we extracted and compared the results and selected the best the best model
depending on the daily
depending on closing price.
the daily closing price.

Figure 1. Methodology
Figure 1. Methodology of processing
of processing data and
data and model model selection.
selection.

AI 2021, 2, FOR PEER REVIEW Figures 2–4 illustrate the training and testing dataset for every targeted cryptocurrency.
6
Figures 2–4 illustrate the training and testing dataset for every targeted cryptocur-
We can see that the price for each currency roughly increased and decreased together along
rency. We can see that the price for each currency roughly increased and decreased to-
the time-series.
gether along the time-series.

Figure 2. Training and testing dataset for BTC.

Figure 2. Training and testing dataset for BTC.

Figure 2 illustrates the BTC closing price within the targeted collected dataset. It
shows that the closing price increased gradually until the end of 2020, when the price
 Figure 2. Training and testing dataset for BTC.

AI 2021, 2 Figure 2 illustrates the BTC closing price within the targeted collected dataset.482
It
shows that the closing price increased gradually until the end of 2020, when the price
increased suddenly, it reach a high of 63,381 USD in a top of peak of time series.

AI 2021, 2, FOR PEER REVIEW 7

Training
Figure 3.and
Figure 3. Training and
testing testing
dataset fordataset
ETH. for ETH.

Figure 3 shows the ETH closing price within the targeted collected dataset. It demon-
strates that the closing price increased gradually until the end of 2020, then the price in-
creased suddenly, reaching a high of 4140 USD.

Figure 4. Training
Figure 4.and testing
Training dataset
and fordataset
testing LTC. for LTC.

Figure 42 shows
illustrates the BTC
the LTC closing
closing priceprice within
within the targeted
the targeted collected
collected dataset.
dataset. It
It illus-
showsthat
trates thatthe
theclosing
closingprice
priceincreased
increasedgradually
graduallyuntil
untilthe
theend
end
ofof 2020,
2020, when
then the the price
price in-
increased
creased suddenly,
suddenly, it reach aa high
reaching high of
of 373.64
63,381 USD.
USD in a top of peak of time series.
Figure 3 shows the ETH closing price within the targeted collected dataset. It demon-
strates
3.1. that the
Machine closing
Learning price increased gradually until the end of 2020, then the price
Algorithms
increased suddenly, reaching
This section demonstrates a high
threeoftypes
4140 of
USD.
machine learning algorithms—long short-
Figure 4 shows the LTC closing price
term memory (LSTM), bidirectional LSTM (bi-LSTM), within the targeted collected
and gated dataset.
recurrent It illustrates
unit (GRU).
that the closing price increased gradually until the end of 2020, then the price increased
suddenly,
3.1.1. Longreaching
Short-Term a high of 373.64
Memory USD.
(LSTM)
For various
3.1. Machine learning
Learning issues involving sequential data, recurrent neural networks with
Algorithms
long short-term memory (LSTM) have emerged as an effective and scalable approach.
This section demonstrates three types of machine learning algorithms—long short-
They are useful for capturing long-term temporal dependencies since they are generic and
term memory (LSTM), bidirectional LSTM (bi-LSTM), and gated recurrent unit (GRU).
effective [43]. The LSTM is an RNN-style architecture with gates that govern the flow of
information between cells.
3.1.1. Long Short-Term The input
Memory (LSTM)and forget gate structures can modify information
traveling along the cell state, with the
For various learning issues involving ultimate output
sequential being
data, a filtered
recurrent version
neural of thewith
networks cell
state
long based on context
short-term memoryfrom(LSTM)
the inputs
have[44]. The LSTM
emerged design
as an has been
effective and criticized for being
scalable approach.
ad hoc and for having a large number of components whose purpose is not
They are useful for capturing long-term temporal dependencies since they are generic and immediately
clear. As [43].
effective a result,
The itLSTM
is unclear whether thearchitecture
is an RNN-style LSTM is the bestgates
with design,
thatand it isthe
govern likely that
flow of
better ones exist [45]. Figure 5 illustrates the structure of a LSTM algorithm
information between cells. The input and forget gate structures can modify information [46].
AI 2021, 2 483

traveling along the cell state, with the ultimate output being a filtered version of the cell
state based on context from the inputs [44]. The LSTM design has been criticized for being
ad hoc and for having a large number of components whose purpose is not immediately
clear. As a result, it is unclear whether the LSTM is the best design, and it is likely that
better ones exist [45]. Figure 5 illustrates the structure of a LSTM algorithm [46].

Figure 5. The structure of a long short-term memory (LSTM) algorithm.

The forward training process of the LSTM can be formulated with the following
equations:  
f t = σ W f · [ h t −1 , x t ] + b f (1)
 
i t = σ W f · [ h t − 1 , x t ] + bi (2)

Ct = f t ∗ Ct−1 + it ∗ tanh(Wc · [ht−1 , xt ] + bc ) (3)

Ot = σ (W0 · [ht−1 , xt ] + b0 ) (4)
ht = ot ∗ tabh(Ct ) (5)
where it , ot , and f t denote the activation of the input gate, output gate, and forget gate,
respectively; Ct and ht denote the activation vector for each cell and memory block, respec-
tively; and W and b denote the weight matrix and bias vector, respectively. In addition,
σ (◦ ) denotes the sigmoid function [47].

3.1.2. Bidirectional LSTM (bi-LSTM)

bi-LSTM was created by Schuster and Paliwal [48] to train a network utilizing past
and future input data sequences. The input data are processed using two linked layers [49].
Bi-directional LSTM predicts or tags the sequence of each element using a finite sequence
based on the context of elements in the past and future. This is the result of two LSTMs
1
AI 2021, 2 484

AI 2021, 2, FOR PEER REVIEW running in parallel, one from left to right and the other from right to left. The forecast of9
a given target signal is known as composite output. This method has proven to be quite
beneficial. The forward function of bi-LSTM with inputs of L units and H as the number
of hidden
number ofunits is calculated
hidden using Equations
units is calculated (6) and (7),
using Equations Figure
(6) and (7),6Figure
shows6ashows
bidirectional
a bidi-
LSTM structure [50]. The hidden layer of the bi-directional LSTM network
rectional LSTM structure [50]. The hidden layer of the bi-directional LSTM network savessaves
two
values. A participates in the forward calculation, and A transpose is involved in
two values. A participates in the forward calculation, and A transpose is involved in thethe reverse
calculation.
reverse The final
calculation. output
The final value,
outputy,value,
depends on A and
y, depends onAAtranspose [51]:
and A transpose [51]:
L H
h ==∑ xl𝑥wlh
𝑤 + + ∑ bht−
t t
a𝑎 𝑏0 1 wh𝑤0 h (6)
l =1 h0 , t>, 0

ath𝑎 ==θh𝜃 a(𝑎

t ) (7)


h (7)

Figure 6. The structure of a bi-directional LSTM (bi-LSTM) algorithm.

Figure 6. The structure of a bi-directional LSTM (bi-LSTM) algorithm.
3.1.3. Gated Recurrent Unit (GRU)
3.1.3.Gated
Gatedrecurrent
Recurrent Unit networks
neural (GRU) (Gated RNNs) have demonstrated their effectiveness
in a Gated
varietyrecurrent neural networks
of applications requiring (Gated RNNs)
sequential have demonstrated
or temporal data [52]. their effective-
The transition
ness in a variety
functions of applications
in hidden units of GRU requiring
are givensequential
as followsor[53]:
temporal data [52]. The transition
functions in hidden units of GRU are given as follows [53]:
z t = σ (W z x t + V z h t − 1 + b z ) (8)
𝑧 = 𝜎(𝑊 𝑥 + 𝑉 ℎ + 𝑏 ) (8)
r t = σ (W r x t + V r h t − 1 + b r ) (9)
𝑟 = 𝜎(𝑊 𝑥 + 𝑉 ℎ + 𝑏 ) (9)
ht = tanh(W c xt + V c (rt · ht−1 )) (10)
ℎ = 𝑡𝑎𝑛ℎ(𝑊 𝑥 + 𝑉 (𝑟 . ℎ )) (10)
h t = (1 − z t ) · h t −1 + z t · h t (11)
ℎ = (1 − 𝑧 ). ℎ + 𝑧 .ℎ (11)
where the model parameters include all W eRdXd , V eRdXd , and b eRdXd that are shared by
where the model parameters include all W𝜖𝑅 , V𝜖𝑅 , and b𝜖𝑅 that are shared by
all time steps and learned during the training stage, (.) denotes the element-wise product,
all time steps and learned during the training stage, (.) denotes the element-wise product,
and k is a hyper-parameter that represents the dimensionality of hidden vectors. Figure 7
and k is a hyper-parameter that represents the dimensionality of hidden vectors. Figure 7
illustrates the GRU structure [54]:
illustrates the GRU structure [54]:
AI 2021, 2 485
AI 2021, 2, FOR PEER REVIEW 10

Figure 7.Figure
The cell
7. model
The cellofmodel
a GRUofblock
a GRUdiagram.
block diagram.

3.2.
3.2. Evaluation
Evaluation Matrix
Matrix
The
The evaluation of
evaluation of the
the proposed
proposed schemes
schemes is
is done
done using
using the
the mean
mean absolute
absolute percentage
error
error (MAPE)
(MAPE) and the root mean squared error (RMSE).
∑ ( )
s
RMSE =∑ N ( Ai − Fi )2 (12)
I =1
RMSE = (12)
| N |
MAPE = ∑ 𝑋 100% (13)
| Ai − Fi |
 
1
N ∑ i=1 cryptocurrency
N
MAPEand
where 𝐴 and 𝐹 are the actual = forecasted × 100% (13)
prices, respectively, and N
Ai
is the number of samples.
where Ai and Fi are the actual and forecasted cryptocurrency prices, respectively, and N is
the Data
3.3. number of samples.
Exploration
When
3.3. Data dealing with data, it is useful and important to understand the data distribu-
Exploration
tion and
When dealingusing
behavior a stable
with data, it is and meaningful
useful charttotounderstand
and important extract thethe
story
datathat the data
distribution
tells. Figure 8 illustrates the time series for targeted cryptocurrency distributed
and behavior using a stable and meaningful chart to extract the story that the data through
tells.
the interval between 22 January 2018 and 30 June 2021. It shows that the price
Figure 8 illustrates the time series for targeted cryptocurrency distributed through the increases
along
intervalwith the specific
between interval
22 January 2018depending on 2021.
and 30 June the closing
It showsprice.
that the price increases along
with the specific interval depending on the closing price.
 AI 2021, 2 486
AI 2021, 2, FOR PEER REVIEW 11

Figure 8. 8.
Figure Time series
Time with
series closing
with prices
closing forfor
prices BTC, ETH,
BTC, and
ETH, LTC.
and LTC.

The
Thecorrelation
correlationmatrix
matrixinin
Figure
Figure9 illustrates the
9 illustrates thecorrelation
correlationcoefficient between
coefficient between the
the
1variables
variables (closing
(closing price).
price). TheThe matrix
matrix shows
shows a strong
a strong positive
positive correlation
correlation between
between differ-
different
ent currencies
currencies (BTC, (BTC, LTC,ETH).
LTC, and and ETH). This means
This means if oneifofone
theoftargeted
the targeted cryptocurrencies
cryptocurrencies in-
increases
creases or decreases,
or decreases, the others
the others behave
behave accordingly.
accordingly.

Figure 9. 9.
Figure Correlation matrix
Correlation forfor
matrix thethe
cryptocurrencies BTC,
cryptocurrencies LTC,
BTC, and
LTC, ETH.
and ETH.

4. 4. Dataset
Dataset
TheThe analyzed
analyzed dataset
dataset was
was collected
collected from
from [42],
[42], ananopen-access
open-access website.
website. It It consists
consists ofof
oneone .csv
.csv file
file separated
separated into
into three
three sheets;
sheets; the
the first
first sheet
sheet forforBitcoin
Bitcoin(BTC),
(BTC),the thesecond
secondforfor
Litecoin
Litecoin (LTC),
(LTC), and
and the
the last
last sheet
sheet forfor Ethereum
Ethereum (ETH).
(ETH). TheThe recorded
recorded prices
prices ininthethe dataset
dataset
werecollected
were collectedon onaadaily
daily basis
basis from
from 11January
January20182018toto3030June
June2021. In In
2021. thisthis
research,
research,we we
used
used time-series data from [19] with 1277 records. Table 1 illustrates the dataset specifica-of
time-series data from [19] with 1277 records. Table 1 illustrates the dataset specification
theoftargeted
tion cryptocurrency
the targeted cryptocurrencyand Figure 10 shows
and Figure sample
10 shows data from
sample data the
from dataset.
the dataset.

Table 1. Dataset specification.

Variable Name Variable Description Data Type

Date Date of Observation Date
AI 2021, 2 487

AI 2021, 2, FOR PEER REVIEW 12

Table 1. Dataset specification.

Variable Name Variable Description Data Type

Opening price on the given
Open
Date Date of Observation Number
Date
day
Open Opening price on the given day Number
High
High HighHigh
priceprice
on the given
on the dayday
given Number
Number
Low
Low Low Low
priceprice
on the given
on the dayday
given Number
Number
Close
Close CloseClose
priceprice on the
on the given
given dayday Number
Number

10. Screenshot
Figure 10.
Figure Screenshotshowing
showinga sample of of
a sample thethe
data from
data the BTC,
from ETH,ETH,
the BTC, and LTC
and dataset.
LTC dataset
5. Results
5. Results
This section shows the results obtained from long short-term memory (LSTM), gated
This section shows the results obtained from long short-term memory (LSTM), gated
recurrent unit (GRU), and bidirectional LSTM (bi-LSTM) algorithms using three types of
recurrent unit (GRU), and bidirectional LSTM (bi-LSTM) algorithms using three types of
popular cryptocurrency: BTC, ETH, and LTC. For each model, the results are illustrated
popular cryptocurrency: BTC, ETH, and LTC. For each model, the results are illustrated
in Tables 2–4. The model that gives the lowest RMSE and MAPE is considered the best
in Tables 2–4. The model that gives the lowest RMSE and MAPE is considered the best
model. Based on this criteria, all of the models applied to three types of currencies can be
model. Based on this criteria, all of the models applied to three types of currencies can be
considered good models but the GRU was found to be the best of the three. The RMSE
considered good models but the GRU was found to be the best of the three. The RMSE of
of the GRU model is the lowest. Thus, GRU is more capable of predicting long-term
the GRU model is the lowest. Thus, GRU is more capable of predicting long-term depend-
dependencies as compared to LSTM and bi-LSTM. This is due to the dependency on past
encies as compared to LSTM and bi-LSTM. This is due to the dependency on past prices.
prices. Figures
Figures 11–19 11–19 illustrate
illustrate the comparisons
the comparisons between thebetween
actual the
andactual and theresults.
the predicted predicted
results. Simulation
Simulation results
results from thosefrom those
models models
indicate indicate
that that
there are there
few are few
occasions occasions
where where
the fore-
the forecast result differs from the actual
cast result differs from the actual results. results.

Table 2. BTC models results.

Table 2. BTC models results.

Model Model RMSE RMSE MAPE MAPE

LSTM LSTM 410.399 410.399 1.1234% 1.1234%
bi-LSTM bi-LSTM 2927.006 2927.006 5.990% 5.990%
GRU 174.129
GRU 174.129 0.2454% 0.2454%

Table 3. ETH models results.

Model RMSE MAPE

LSTM 59.507 1.5498%
bi-LSTM 321.061 6.85%
GRU 26.59 0.8267%
AI 2021, 2 488

Table 3. ETH models results.

Model RMSE MAPE

LSTM 59.507 1.5498%
bi-LSTM 321.061 6.85%
GRU 26.59 0.8267%

Table 4. LTC model results.

Model RMSE MAPE

LSTM 3.069 0.8474%
bi-LSTM 4.307 2.332%
GRU 0.825 0.2116%

Figure 11. Actual and predicted price of BTC using the LSTM model.

Figure 12. Actual and predicted price of BTC using the GRU model.
AI
AI2021,
AI 2021,2,2,
2021, 2FOR
FORPEER
PEERREVIEW
REVIEW 14
489
14
AI 2021, 2, FOR PEER REVIEW 14

Figure13.
Figure
Figure 13. Actualand
13.Actual
Actual andpredicted
and predictedprice
predicted priceof
price ofBTC
of BTCusing
BTC usingthe
using thebi-LSTM
the bi-LSTMmodel.
bi-LSTM model.
model.
Figure 13. Actual and predicted price of BTC using the bi-LSTM model.

Figure
Figure14.
Figure 14.Actual
14. Actualand
Actual andpredicted
and predictedprice
predicted priceof
price ofETH
of ETHusing
ETH usingthe
using theLSTM
the LSTMmodel.
LSTM model.
model.
Figure 14. Actual and predicted price of ETH using the LSTM model.

Figure
Figure15.
Figure 15.Actual
15. Actualand
Actual andpredicted
and predictedprice
predicted priceof
price ofETH
of ETHusing
ETH usingthe
using theGRU
the GRUmodel.
GRU model.
model.
Figure 15. Actual and predicted price of ETH using the GRU model.
AI
AI
AI 2021,
2021, 2,
22, FOR PEER REVIEW 15
490
AI 2021,
2021, 2, FOR
FOR PEER
PEER REVIEW
REVIEW 15
15

Figure
Figure 16.
Figure
Figure 16. Actual
16.
16. Actualand
Actual
Actual and predicted
and
and predicted price
predicted
predicted price of
price
price of ETH
of
of ETH using
ETH
ETH using the
using
using the bi-LSTM
the
the bi-LSTM model.
bi-LSTM
bi-LSTM model.
model.
model.

Figure
Figure 17.
Figure
Figure 17. Actual
17.
17. Actualand
Actual
Actual and predicted
and
and predicted price
predicted
predicted price of
price
price of LTC
of
of LTC using
LTC
LTC using the
using the LSTM
the LSTM model.
LSTM model.
model.

Figure 18. Actual and predicted price of LTC using the GRU model.
Figure
Figure 18.
Figure 18. Actual
18. Actual and
Actual and predicted
and predicted price
predicted price of
price of LTC
of LTC using
LTC using the
using the GRU
the GRU model.
GRU model.
model.
AI 2021,
2021, 22, FOR PEER REVIEW 491
16

Figure 19. Actual and predicted price of LTC using the bi-LSTM model.
Figure 19. Actual and predicted price of LTC using the bi-LSTM model.
5.1. Results for BTC
5.1. Results for BTC of these models for BTC cryptocurrency are tabulated in Table 2.
The accuracies
The MAPE of the GRU
The accuracies model
of these modelsis the
for lowest with a value are
BTC cryptocurrency of 0.2454
tabulatedandinthe RMSE
Table is
2. The
174.129.
MAPE ofTherefore,
the GRU GRU model is is
more
the capable of predicting
lowest with a value of BTC trends
0.2454 and than
theLSTM
RMSEorisbi-LSTM,
174.129.
with a small
Therefore, GRU difference
is more between
capable ofit predicting
and the LSTM model.than
BTC trends Figures
LSTM 11–13 show a with
or bi-LSTM, visuala
representation,
small differencecomparing
between it the andactual
the LSTMand predicted values11–13
model. Figures of theshow
training dataset
a visual of the
represen-
three models for BTC.
tation, comparing the actual and predicted values of the training dataset of the three mod-
els forResults
BTC. presented in Figure 11 compare the actual and LSTM-predicted price of BTC.
The graph shows that the predicted and the actual price is approximately the same over
the entire
Resultsinterval.
presented Thisinmodel
Figureis11considered
compare the theactual
second-best model. The mean
and LSTM-predicted priceabsolute
of BTC.
percentage
The error for
graph shows the
that theprediction
predictedmodel
and the(MAPE)
actual of BTC
price is for LSTM is 1.1234%,
approximately the sameandover
the
rootentire
the mean interval.
square errorThis(RMSE)
model is is 410.399.
consideredStatistical analysis ofmodel.
the second-best the data indicates
The that the
mean absolute
predicted price
percentage errorhasfora the
mean value of model
prediction 38,173.258 USD,of
(MAPE) a maximum
BTC for LSTM valueisof1.1234%,
64,358.805andUSD,
the
and a minimum value of 12,775.013 USD, whereas the actual price has
root mean square error (RMSE) is 410.399. Statistical analysis of the data indicates that the a mean value of
38,249.388 USD, a maximum value of 63,380.999 USD, and a minimum
predicted price has a mean value of 38,173.258 USD, a maximum value of 64,358.805 USD, value of 12,941.0
USD.a The
and mean difference
minimum between theUSD,
value of 12,775.013 meanwhereas
values of theactual
the actualprice
and thehas predicated
a mean value prices
of
is 76.13 USD.
38,249.388 USD, a maximum value of 63,380.999 USD, and a minimum value of 12,941.0
USD.Figure
The mean12 illustrates
differencethe comparison
between thebetween the actual
mean values andactual
of the the predicted
and theprice of BTC
predicated
using the GRU
prices is 76.13 USD.model. The graph shows that the difference between the predicted and the
actualFigure 12 illustrates the comparison between the actual and the predicted price in
price is virtually non-existent along the testing set, with very small differences of
the top
BTC using fewthe peaks
GRU of model.
the time series.
The graphThis model
shows is the
that considered
difference to between
be the best.
the The mean
predicted
absolute
and percentage
the actual price iserror for thenon-existent
virtually prediction model (MAPE)
along the testingof set,
the GRU model
with very for BTC
small is
differ-
0.2454%, and the root mean square error (RMSE) is 174.29. Statistical analysis of the data
ences in the top few peaks of the time series. This model is considered to be the best. The
indicates that the predicted price has a mean value of 38,333.36 USD, a maximum value
mean absolute percentage error for the prediction model (MAPE) of the GRU model for
of 63,870.605 USD, and a minimum value of 12,936.112 USD, whereas the actual price has
BTC is 0.2454%, and the root mean square error (RMSE) is 174.29. Statistical analysis of
a mean value of 38,249.388 USD, a maximum value of 63,380.999 USD, and a minimum
the data indicates that the predicted price has a mean value of 38,333.36 USD, a maximum
value of 12,941.0 USD. The mean difference between the mean values of the actual and the
value of 63,870.605 USD, and a minimum value of 12,936.112 USD, whereas the actual
predicated prices is 83.97 USD.
price has a mean value of 38,249.388 USD, a maximum value of 63,380.999 USD, and a
The results in Figure 13 illustrate the comparison between the actual and the predicted
minimum value of 12,941.0 USD. The mean difference between the mean values of the
price of BTC under the bi-LSTM model. It indicates a greater difference between the actual
actual and the predicated prices is 83.97 USD.
and the predicted price compared with the LSTM and GRU models. The mean absolute
The results in Figure 13 illustrate the comparison between the actual and the pre-
percentage error prediction model (MAPE) from bi-LSTM for BTC is 5.990% and the root
dicted price of BTC under the bi-LSTM model. It indicates a greater difference between
mean square error (RMSE) is 2927.006. Statistical analysis of the data indicates that the
the actual price
predicted and the haspredicted
a mean valuepriceofcompared
36,735.137with
USD, thea LSTM
maximum and value
GRU models. The mean
of 59,885.746 USD,
absolute
and a minimum value of 12,941.0 USD, whereas the actual price has a mean valueand
percentage error prediction model (MAPE) from bi-LSTM for BTC is 5.990% of
the root mean
38,249.388 USD,square error (RMSE)
a maximum value ofis63,380.999
2927.006.USD,
Statistical
and a analysis
minimum ofvalue
the data indicates
of 13,655.032
AI 2021, 2 492

USD. The mean differences between the mean values of the actual and the predicated
prices is 1514.251 USD.

5.2. Results for ETH

The accuracies of these models for ETH cryptocurrency are tabulated in Table 3. The
mean absolute percentage error for the GRU model is the least with a value of 0.8267
and a root mean square error of 26.59. Therefore, GRU proved to be the best predictor
compared to LSTM and bi-LSTM for ETH. Figures 14–16 show the visual representation of
the comparison between the actual and the predicted values of the training dataset of the
three models for the ETH.
LSTM model for ETH, it represents that the difference between the predicted and
the actual price is very small as red and green curves moving over each other’s over the
whole period of time of Figure 14. This model is considered the second-best model. The
mean absolute percentage error prediction model of ETH for the LSTM model is 1.5489%
and the root mean square error is 59.507. Statistical analysis of the data indicates that the
predicted price has a mean value of 1663.1392 USD, a maximum value of 4399.33 USD, and
a minimum value of 379.41837 USD, whereas the actual price has a mean value of 1636.7091
USD, a maximum value of 4140.0 USD, and a minimum value of 383.35 USD. The mean
difference between the mean values of the actual and the predicated prices is 26.43 USD.
Figure 15 illustrates the comparison between the actual and the predicted price of the
GRU model for ETH. It represents a negligible difference between the predicted and the
actual price along the testing set of the time series. This model is considered the best with a
mean absolute percentage of 0.8267%, and root mean square error of 321.061. Statistical
analysis of the data indicates that the predicted price has a mean value of 1655.4645 USD, a
maximum value of 4249.46 USD, and a minimum value of 384.33 USD, whereas the actual
price has a mean value of 1636.7091 USD, a maximum value of 4140.0 USD, and a minimum
value of 383.35 USD. The mean difference between the mean values of the actual and the
predicated prices is 18.76 USD.
The results in Figure 16 illustrate the comparison between the actual and the predicted
price of the bi-LSTM model for ETH. It shows substantial differences between the actual
and the predicted price compared with the LSTM and GRU models with mean absolute
percentage error of 6.85% and root mean square error of 321.061. Statistical analysis of the
data indicates that the predicted price has a mean value of 1733.5935 USD, a maximum
value of 4576.371 USD, and a minimum value of 350.24 USD, whereas the actual price has
a mean value of 1636.7091 USD, a maximum value of 4140.0 USD, and a minimum value of
383.35 USD. The mean difference between the mean values of the actual and the predicated
prices is 96.88 USD.

5.3. Results for LTC

The accuracy of the models for the LTC cryptocurrency are shown in Table 4. The
mean absolute percentage error of the GRU model is the lowest with a value of 0.2116 and
a root mean square error of 0.825. Therefore, GRU proved to be most capable for prediction
as compared to LSTM and bi-LSTM for LTC. Figures 17–19 show the visual representation
of the data by comparing the actual and the predicted values of training dataset of the three
models for LTC.
The results in Figure 17 show the comparison between the actual and the predicted
price of the LSTM model for LTC. They show that the difference between the predicted
and the actual price is very small with a mean absolute percentage error of 0.8474%, and a
root mean square error of 3.069. Statistical analysis of the data indicates that the predicted
price has a mean value of 166.16 USD, a maximum value of 388.59 USD, and a minimum
value of 53.95 USD, whereas the actual price has a mean value of 165.68 USD, a maximum
value of 373.64 USD, and a minimum value of 53.64 USD. The mean difference between
the mean values of the actual and the predicated prices is 0.48 USD.
AI 2021, 2 493

Figure 18 shows the comparison between the actual and the predicted price of the GRU
model for LTC. It represents similar trends for both actual and predicted price with a mean
absolute percentage error of 0.2116% and a root mean square error of 0.825. This model
is considered the best model. Statistical analysis of the data indicates that the predicted
price has a mean value of 165.93 USD, a maximum value of 378.10 USD, and a minimum
value of 53.84 USD, whereas the actual price has a mean value of 165.68 USD, a maximum
value of 373.64 USD, and a minimum value of 53.64 USD. The mean difference between
the mean values of the actual and the predicated prices is 0.25 USD.
Figure 19 illustrates the comparison between the actual and the predicted price of
the bi-LSTM model for LTC. It represents a small difference between the actual and the
predicted price with mean absolute percentage error of 2.332% and a root mean square
error of 4.307. Statistical analysis of the data indicates that the predicted price has a mean
value of 176.75 USD, a maximum value of 378.10 USD, and a minimum value of 51.12 USD,
whereas the actual price has a mean value of 165.68 USD, a maximum value of 397.57 USD,
and a minimum value of 53.64 USD. The mean difference between the mean values of the
actual and the predicated prices is 11.07 USD.

6. Discussion
The proposed model in this research can be considered a reliable and acceptable model
for cryptocurrency prediction. Table 5 presents a comparison between the proposed model
in this paper and other models in the literature.

Table 5. A comparison between previous work and the model proposed in this paper.

Reference Cryptocurrency Method Result

[51] BTC, LTC Multi-linear regression model R2 score: 44% for LTC and 59% for BTC
Logistic regression and linear LR: 66%
[45] BTC
discriminant analysis LDA: 65.3%
RMSE ARIMA: 302.53, LSTM: 603.68
[53] BTC ARIMA, LSTM and GRU.
GRU: 381.34
GRU
MAPE: 0.2454%
RMSE: 174.129
LSTM
This paper BTC LSTM, GRU, and bi-LSTM MAPE: 1.1234%
RMSE: 410.399
bi-LSTM
MAPE: 5.990%
RMSE: 2927.006
GRU
MAPE: 0.8267%
RMSE: 26.59
LSTM
This paper ETH LSTM, GRU, and bi-LSTM MAPE: 1.5498%
RMSE: 59.507
bi-LSTM
MAPE: 6.85%
RMSE: 321.061
GRU
MAPE: 0.2116%
RMSE: 0.825
LSTM
This paper LTC LSTM, GRU, and bi-LSTM MAPE: 0.8474%
RMSE: 3.069
bi-LSTM
MAPE: 2.332%
RMSE: 4.307
AI 2021, 2 494

Table 5 shows a comparison between the proposed model in this paper and other
models in the literature [45,51,53]. The MAPES values of the proposed model in this
paper for GRU predicting LTC represents the best performance compared to all other
models as the predicted results are very close to the actual results. Results obtained from
this paper show that the GRU performed better when predicting the price of all types of
cryptocurrency than the LSTM and the bi-LSTM models.

7. Conclusions
In this paper, three types of machine learning algorithm are constructed and used for
predicting the prices of three types of cryptocurrency—BTC, ETH, and LTC. Performance
measures were conducted to test the accuracy of different models as shown in Tables 2–4.
Then, we compared the actual and predicted prices. The results show that GRU outper-
formed the other algorithms with a MAPE of 0.2454%, 0.8267%, and 0.2116% for BTC, ETH,
and LTC, respectively. The RMSE for the GRU model was found to be 174.129, 26.59, and
0.825 for BTC, ETH, and LTC, respectively. Based on these outcomes, the GRU model
for the targeted cryptocurrencies can be considered efficient and reliable. This model is
considered the best model. However, bi-LSTM represents less accuracy than GRU and
LSTM with substantial differences between the actual and the predicted prices for both
BTC and ETH. The experimental results show that:
• The AI algorithm is reliable and acceptable for cryptocurrency prediction.
• GRU can predict cryptocurrency prices better than LSTM and bi-LSTM but overall all
algorithms represent excellent predictive results.
In future work, we will investigate other factors that might affect the prices of the
cryptocurrency market, and we will focus on the effect that social media in general and
tweets in particular can have on the price and trading volume of cryptocurrencies by
analyzing tweets using natural language processing techniques and sentiment analysis.

Author Contributions: M.J.H. designed and built the models, processed the data, and wrote the first
draft of the manuscript; A.Y.O. helped in the processing of the data and in the interpretation of the
results. A.Y.O. revised and edited the manuscript. All authors have read and agreed to the published
version of the manuscript.
Funding: This research is funded by the Arab American University in Palestine.
Data Availability Statement: Data used for this article is publicly available and collected from
https://www.marketwatch.com.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Mukhopadhyay, U.; Skjellum, A.; Hambolu, O.; Oakley, J.; Yu, L.; Brooks, R. A brief survey of Cryptocurrency systems. In
Proceedings of the 14th Annual Conference on Privacy, Security and Trust (PST), Auckland, New Zealand, 12–14 December 2016;
pp. 745–752. [CrossRef]
2. Rose, C. The Evolution Of Digital Currencies: Bitcoin, A Cryptocurrency Causing A Monetary Revolution. Int. Bus. Econ. Res. J.
(IBER) 2015, 14, 617. [CrossRef]
3. Brenig, C.; Accorsi, R.; Müller, G. Economic Analysis of Cryptocurrency Backed Money Laundering. ECIS 2015 Completed
Research Papers. Paper 20. 2015. Available online: https://aisel.aisnet.org/ecis2015_cr/20 (accessed on 16 June 2021).
4. Eyal, I. Blockchain Technology: Transforming Libertarian Cryptocurrency Dreams to Finance and Banking Realities. Computer
2017, 50, 38–49. [CrossRef]
5. DeVries, P. An Analysis of Cryptocurrency, Bitcoin, and the Future. Int. J. Bus. Manag. Commer. 2016, 1, 1–9.
6. Jang, H.; Lee, J. An Empirical Study on Modeling and Prediction of Bitcoin Prices with Bayesian Neural Networks Based on
Blockchain Information. IEEE Access 2017, 6, 5427–5437. [CrossRef]
7. Saad, M.; Choi, J.; Nyang, D.; Kim, J.; Mohaisen, A. Toward Characterizing Blockchain-Based Cryptocurrencies for Highly
Accurate Predictions. IEEE Syst. J. 2019, 14, 321–332. [CrossRef]
8. Gautam, K.; Sharma, N.; Kumar, P. Empirical Analysis of Current Cryptocurrencies in Different Aspects. In Proceedings of the
ICRITO 2020—IEEE 8th International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future
Directions), Noida, India, 4–5 June 2020; pp. 344–348. [CrossRef]
AI 2021, 2 495

9. Adams, R.; Kewell, B.; Parry, G. Blockchain for Good? Digital Ledger Technology and Sustainable Development Goals. In
Handbook of Sustainability and Social Science Research; Filho, W.L., Marans, R., Callewaert, J., Eds.; World Sustainability Series;
Springer: Cham, Switzerland, 2018. [CrossRef]
10. Killer, C.; Rodrigues, B.; Stiller, B. Security Management and Visualization in a Blockchain-based Collaborative Defense. In
Proceedings of the ICBC 2019—IEEE International Conference on Blockchain and Cryptocurrency, Seoul, Korea, 14–17 May 2019;
pp. 108–111. [CrossRef]
11. Gandal, N.; Halaburda, H. Competition in the Cryptocurrency Market (September 29, 2014). CESifo Working Paper Series No.
4980. Available online: https://ssrn.com/abstract=2506577 (accessed on 16 June 2021).
12. Iwamura, M.; Kitamura, Y.; Matsumoto, T. Is Bitcoin the Only Cryptocurrency in the Town? Economics of Cryptocurrency
And Friedrich A. Hayek (February 28, 2014). Available online: https://ssrn.com/abstract=2405790 (accessed on 16 June 2021).
[CrossRef]
13. Kyriazis, N.A. A Survey on Empirical Findings about Spillovers in Cryptocurrency Markets. J. Risk Financ. Manag. 2019, 12, 170.
[CrossRef]
14. Hassani, H.; Huang, X.; Silva, E. Big-Crypto: Big Data, Blockchain and Cryptocurrency. Big Data Cogn. Comput. 2018, 2, 34.
[CrossRef]
15. Nizzoli, L.; Tardelli, S.; Avvenuti, M.; Cresci, S.; Tesconi, M.; Ferrara, E. Charting the Landscape of Online Cryptocurrency
Manipulation. IEEE Access 2020, 8, 113230–113245. [CrossRef]
16. Rebane, J.; Karlsson, I.; Papapetrou, P.; Denic, S. Seq2Seq RNNs and ARIMA models for Cryptocurrency Prediction: A Compara-
tive Study. In Proceedings of the SIGKDD Workshop on Fintech (SIGKDD Fintech’18), London, UK, 19–23 August 2018.
17. Rehman, M.U.; Apergis, N. Determining the predictive power between cryptocurrencies and real time commodity futures:
Evidence from quantile causality tests. Resour. Policy 2019, 61, 603–616. [CrossRef]
18. Liew, J.; Li, R.Z.; Budavári, T.; Sharma, A. Cryptocurrency Investing Examined. J. Br. Blockchain Assoc. 2019, 2, 1–12. [CrossRef]
19. Dyntu, V.; Dykyi, O. Cryptocurrency in the system of money laundering. Balt. J. Econ. Stud. 2019, 4, 75–81. [CrossRef]
20. Kethineni, S.; Cao, Y. The Rise in Popularity of Cryptocurrency and Associated Criminal Activity. Int. Crim. Justice Rev. 2019, 30,
325–344. [CrossRef]
21. Liu, Y.; Tsyvinski, A. Risks and Returns of Cryptocurrency. Rev. Financ. Stud. 2020, 34, 2689–2727. [CrossRef]
22. Valdeolmillos, D.; Mezquita, Y.; González-Briones, A.; Prieto, J.; Corchado, J.M. Blockchain Technology: A Review of the Current
Challenges of Cryptocurrency. In Blockchain and Applications. BLOCKCHAIN 2019. Advances in Intelligent Systems and Computing;
Prieto, J., Das, A., Ferretti, S., Pinto, A., Corchado, J., Eds.; Springer: Cham, Switzerland, 2020; Volume 1010. [CrossRef]
23. Yuneline, M.H. Analysis of cryptocurrency’s characteristics in four perspectives. J. Asian Bus. Econ. Stud. 2019, 26, 206–219.
[CrossRef]
24. Huynh, T.L.D.; Nasir, M.A.; Vo, X.V.; Nguyen, T.T. “Small things matter most”: The spillover effects in the cryptocurrency market
and gold as a silver bullet. North Am. J. Econ. Financ. 2020, 54, 101277. [CrossRef]
25. Hitam, N.A.; Ismail, A.R. Comparative Performance of Machine Learning Algorithms for Cryptocurrency Forecasting. Indones. J.
Electr. Eng. Comput. Sci. 2018, 11, 1121–1128. [CrossRef]
26. Andrianto, Y. The Effect of Cryptocurrency on Investment Portfolio Effectiveness. J. Financ. Account. 2017, 5, 229. [CrossRef]
27. Derbentsev, V.; Babenko, V.; Khrustalev, K.; Obruch, H.; Khrustalova, S. Comparative Performance of Machine Learning Ensemble
Algorithms for Forecasting Cryptocurrency Prices. Int. J. Eng. Trans. A Basics 2021, 34, 140–148. [CrossRef]
28. Patel, M.M.; Tanwar, S.; Gupta, R.; Kumar, N. A Deep Learning-based Cryptocurrency Price Prediction Scheme for Financial
Institutions. J. Inf. Secur. Appl. 2020, 55, 102583. [CrossRef]
29. Miura, R.; Pichl, L.; Kaizoji, T. Artificial Neural Networks for Realized Volatility Prediction in Cryptocurrency Time Series. In
Advances in Neural Networks—ISNN 2019; Lu, H., Tang, H., Wang, Z., Eds.; Lecture Notes in Computer Science; Springer: Cham,
Switzerland, 2019; Volume 11554. [CrossRef]
30. Karasu, S.; Altan, A.; Sarac, Z.; Hacioglu, R. Prediction of Bitcoin prices with machine learning methods using time series data. In
Proceedings of the 26th Signal Processing and Communications Applications Conference (SIU), Izmir, Turkey, 2–5 May 2018.
[CrossRef]
31. Saad, M.; Mohaisen, A. Towards characterizing blockchain-based cryptocurrencies for highly-accurate predictions. In Proceedings
of the IEEE INFOCOM—IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS), Honolulu, HI,
USA, 15–19 April 2018. [CrossRef]
32. Yiying, W.; Yeze, Z. Cryptocurrency Price Analysis with Artificial Intelligence. In Proceedings of the 5th International Conference
on Information Management (ICIM), Cambridge, UK, 24–27 March 2019; pp. 97–101. [CrossRef]
33. Chen, Z.; Li, C.; Sun, W. Bitcoin price prediction using machine learning: An approach to sample dimension engineering. J.
Comput. Appl. Math. 2019, 365, 112395. [CrossRef]
34. Valencia, F.; Gómez-Espinosa, A.; Valdés-Aguirre, B. Price Movement Prediction of Cryptocurrencies Using Sentiment Analysis
and Machine Learning. Entropy 2019, 21, 589. [CrossRef]
35. Ferdiansyah, F.; Othman, S.H.; Radzi, R.Z.R.M.; Stiawan, D.; Sazaki, Y.; Ependi, U. A LSTM-Method for Bitcoin Price Prediction: A
Case Study Yahoo Finance Stock Market. In Proceedings of the ICECOS—3rd International Conference on Electrical Engineering
and Computer Science, Batam, Indonesia, 2–3 October 2019; pp. 206–210. [CrossRef]
AI 2021, 2 496

36. Zhao, Y.; Chen, Z. Forecasting stock price movement: New evidence from a novel hybrid deep learning model. J. Asian Bus. Econ.
Studies 2021. ahead-of-print. [CrossRef]
37. Jain, A.; Tripathi, S.; Dwivedi, H.D.; Saxena, P. Forecasting Price of Cryptocurrencies Using Tweets Sentiment Analysis. In
Proceedings of the 11th International Conference on Contemporary Computing (IC3), Noida, India, 2–4 August 2018; pp. 1–7.
[CrossRef]
38. Wu, C.-H.; Lu, C.-C.; Ma, Y.-F.; Lu, R.-S. A New Forecasting Framework for Bitcoin Price with LSTM. In Proceedings of the IEEE
International Conference on Data Mining Workshops (ICDMW), Singapore, 17–20 November 2018; pp. 168–175. [CrossRef]
39. Yamak, P.T.; Yujian, L.; Gadosey, P.K. A Comparison between ARIMA, LSTM, and GRU for Time Series Forecasting. In Proceedings
of the 2nd International Conference on Algorithms, Computing and Artificial Intelligence, Sanya, China, 20–22 December 2019;
pp. 49–55. [CrossRef]
40. McNally, S.; Roche, J.; Caton, S. Predicting the Price of Bitcoin Using Machine Learning. In Proceedings of the 26th Euromicro
International Conference on Parallel, Distributed and Network-based Processing (PDP), Cambridge, UK, 21–23 March 2018;
pp. 339–343. [CrossRef]
41. Lazo, J.G.L.; Medina, G.H.H.; Guevara, A.V.; Talavera, A.; Otero, A.N.; Cordova, E.A. Support System to Investment Management
in Cryptocurrencies. In Proceedings of the 2019 7th International Engineering, Sciences and Technology Conference, IESTEC,
Panama, Panama, 9–11 October 2019; pp. 376–381. [CrossRef]
42. Market Watch Which Provides the Latest Stock Market, Financial and Business News. Available online: https://www.
marketwatch.com (accessed on 16 June 2021).
43. Greff, K.; Srivastava, R.K.; Koutnik, J.; Steunebrink, B.R.; Schmidhuber, J. LSTM: A Search Space Odyssey. IEEE Trans. Neural
Netw. Learn. Syst. 2017, 28, 2222–2232. [CrossRef]
44. Sherratt, F.; Plummer, A.; Iravani, P. Understanding LSTM Network Behaviour of IMU-Based Locomotion Mode Recognition for
Applications in Prostheses and Wearables. Sensors 2021, 21, 1264. [CrossRef] [PubMed]
45. Jozefowicz, R.; Zaremba, W.; Sutskever, I. An empirical exploration of Recurrent Network architectures. In Proceedings of the
32nd International Conference on Machine Learning (ICML), Lille, France, 6–11 July 2015; Volume 3, pp. 2332–2340.
46. Le, X.-H.; Ho, H.V.; Lee, G.; Jung, S. Application of Long Short-Term Memory (LSTM) Neural Network for Flood Forecasting.
Water 2019, 11, 1387. [CrossRef]
47. Li, X.; Peng, L.; Yao, X.; Cui, S.; Hu, Y.; You, C.; Chi, T. Long short-term memory neural network for air pollutant concentration
predictions: Method development and evaluation. Environ. Pollut. 2017, 231, 997–1004. [CrossRef]
48. Kwak, G.H.; Park, C.W.; Ahn, H.Y.; Na, S.; Il, L.; Do, K.; Park, N.W. Potential of bidirectional long short-term memory networks
for crop classification with multitemporal remote sensing images. Korean J. Remote. Sens. 2020, 36, 515–525. [CrossRef]
49. Althelaya, K.A.; El-Alfy, E.-S.M.; Mohammed, S. Evaluation of bidirectional LSTM for short-and long-term stock market prediction.
In Proceedings of the 9th International Conference on Information and Communication Systems (ICICS), Irbid, Jordan., 3–5 April
2018. [CrossRef]
50. Yildirim, Ö. A novel wavelet sequence based on deep bidirectional LSTM network model for ECG signal classification. Comput.
Biol. Med. 2018, 96, 189–202. [CrossRef] [PubMed]
51. Yang, S. Research on Network Behavior Anomaly Analysis Based on Bidirectional LSTM. In Proceedings of the IEEE 3rd
Information Technology, Networking, Electronic and Automation Control Conference (ITNEC), Chengdu, China, 15–17 March
2019. [CrossRef]
52. Dey, R.; Salemt, F.M. Gate-variants of Gated Recurrent Unit (GRU) neural networks. In Proceedings of the IEEE 60th International
Midwest Symposium on Circuits and Systems (MWSCAS), Boston, MA, USA, 6–9 August 2017. [CrossRef]
53. Zhao, R.; Wang, D.; Yan, R.; Mao, K.; Shen, F.; Wang, J. Machine Health Monitoring Using Local Feature-Based Gated Recurrent
Unit Networks. IEEE Trans. Ind. Electron. 2018, 65, 1539–1548. [CrossRef]
54. Su, Y.; Kuo, C.-C.J. On extended long short-term memory and dependent bidirectional recurrent neural network. Neurocomputing
2019, 356, 151–161. [CrossRef]