See discussions, stats, and author profiles for this publication at: [Link]

net/publication/383415689

Impact of Artificial Intelligence (AI) on Stock Market: A comprehensive

systematic review

Article · August 2024

CITATIONS READS

0 1,374

4 authors, including:

Mohamad Ahadzadeh Mahdi Karimi

Islamic Azad University, Science and Research Branch Shahid Beheshti University
9 PUBLICATIONS 1 CITATION 8 PUBLICATIONS 0 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mohamad Ahadzadeh on 26 August 2024.

The user has requested enhancement of the downloaded file.

Impact of Artificial Intelligence (AI) on Stock Market:
A comprehensive systematic review
Mohamad Ahadzadeh1*, Mahdi Karimi2, Ghazaleh Asgari3, Negar Asgari4
1Master Of Financial Management, Faculty Of Management And Economic, Science And Research University, Tehran, Iran.
[Link]@[Link] (Corresponding author).
2Master Of Financial Management, Faculty Of Management And Economic, shahid beheshti University, Tehran, Iran.
mehwolo@[Link]
3bachelor of accounting, Faculty Of Management And Economic, Science And Research University, Tehran, Iran.
asgsriiiqazale1700@[Link]
4bachelor of accounting, Faculty Of Management And Economic, Science And Research University, Tehran, Iran.
negarasgarii78899@[Link]

Abstract
The capabilities of AI in analyzing financial data have led to several studies. in this regard; This
piece delves into a thorough and systematic examination of scholarly research on the application
of AI in the realm of stock market investment for market participants, compiled from a selection
of 2053 articles retrieved from the Web of Science (WOS) database. The literature review
delineates five pivotal research domains where AI has significantly contributed; encompassing the
prediction of stock market trends and volatilities, risk management, portfolio optimization, and
investment optimization. This comprehensive analysis underscores the multifaceted role of AI in
the financial territory, highlighting its potential to revamp traditional financial practices and
decision-making processes. In addition, this review underscores the necessity for additional
research to elucidate unresolved inquiries and comprehend the wider implications of disruptive
technological advancements within the financial sector. Specifically, it discusses five principal
questions concerning the impact of AI on the stock market. This systematic review serves as a
pivotal resource for scholars, offering a comprehensive examination of the contemporary
landscape of AI applications within the financial sector.

Keywords: Artificial Intelligence, Stock Market, Prediction, Optimization, Management

JEL Classifcation: G1, C6, E71
[Link]
In the rapidly evolving global stock markets, the advent of AI has spring up as a transformative
force, significantly altering the dynamics of investment, predictability, portfolio management, risk
management, and market analysis. This systematic review aims to Inquire into the profound impact
of AI on financial markets, providing a comprehensive analysis of the existing literature. The
objective is to uncover the multifaceted applications of AI in finance, shedding light on its role in
enhancing market predictability, stability, and efficiency. In examining the Impact of AI on the
stock market, a multitude of comprehensive studies have been undertaken across various academic
disciplines. (Abakah et al. 2023; Najem et al. 2024; Lin, and Marques 2024;
Jareño and Yousaf 2023; Gonzales and Hargreaves 2022).
Predicting stock market movements presents a formidable yet indispensable challenge for
investors, traders, and researchers alike. A myriad of methodologies, encompassing Time-series
analysis, Predictive modeling, Algorithmic strategies, and AI techniques, have been proposed to
Forecast stock prices and Management & optimization of portfolio, current risks, Improving
decision-making performance in the direction of appropriate investment and surpass market
performance benchmarks. Notably, AI techniques, particularly ML and DL have attracted
significant attention due to their potential to enhance prediction accuracy and efficiency. (Chaajer
et al. 2022; Samitas et al. 2020; Song and Jain 2022). Another pivotal and formidable area for
investors involves the utilization of AI and its methodologies, such as DL and ML, to enhance the
precision and efficiency of risk management and portfolio management within the stock market. a
multitude of research has been carried out on this topic, underscoring the significance of AI in
optimizing investment strategies. (Gonzales and Hargreaves 2022; Jang and Seong 2023;
Rubesam 2022; Mita and Takahashi 2023). Artificial intelligence (AI) is revolutionizing the stock
market by providing advanced review algorithms and models that can review and analyze large
amounts of data and information quickly, accurately and efficiently, enabling investors to make
more informed decisions in their transactions. AI algorithms can identify patterns, trends, and
anomalies in stock market data that human analysts may overlook. This technology can help
investors optimize their portfolios, manage risks, and identify profitable opportunities in real time.
The use of AI in investment strategies has the potential to enhance returns and mitigate losses by
leveraging advanced predictive analytics and ML techniques. As AI continues to evolve, its impact
on the dynamics of investment in the stock market is Predicted to expand, shaping a new era of
data-driven decision-making in finance.
This paper is dedicated to the meticulous analysis and synthesis of scholarly articles within the
realm of business and economics, with a specific focus on the effect of Artificial Intelligence (AI)
on the stock market. Our objective is to pinpoint the pivotal nodes, Such as the most leading articles
and journals, within the relevant research landscape. Furthermore, we endeavor to discern the
predominant research themes concerning the application of AI in the stock market within our
discipline. In addition to this, we intend to propose recommendations for future research efforts
and offer practical insights for businesses interested in implementing AI strategies within the stock
market. In a similar vein, this paper embarks on a systematic and objective review, anchored in
Statistical review. Initially, we offer an all-round overview of the total count and disciplinary
distribution of scholarly articles related to the impact of Artificial Intelligence (AI) on the Stock
Market. A total of 862 Scientific papers were meticulously retrieved for this analysis.
Subsequently, the focus narrows to the domain of business and economics, enriching our dataset
with an additional 234 articles for a more targeted examination. Following this refinement, we
delve into the analysis of influential countries, publications, papers, and the most frequently
occurring keywords within the selected literature. Utilizing a Academic literature review tool, we
successfully identified five core research themes related to the effect of AI on the Stock Market.
This data-driven approach is poised to offer a more objective and nuanced presentation of the
current condition of research in AI field in stock market.
The succeeding structure of this paper is as follows. The subsequent section provides a
comprehensive overview of the extant literature across all disciplines, meticulously detailing the
corpus of papers pertinent to the effect of AI on Stock Markets. This includes a holistic description
of the literature's distribution by discipline, the volume of research, and the geographical and
publication-based distribution of the literature. Following this, a more focused analysis is
conducted within the domain of business and economics, examining the geographical distribution
of publications, the prominence of certain countries in the discourse, and the identification of
highly cited papers within this field. Additionally, the paper outlines the primary research themes,
as identified through VOSviewer, a tool for analyzing citation networks. The paper then delineates
promising research directions and potential practical applications, aiming to stimulate further
exploration and innovation in the field. Concluding the paper, the final section encapsulates the
conclusions drawn from the analysis and acknowledges the limitations encountered during the
research process. This structured approach ensures a thorough examination of the literature,
providing a robust foundation for future research and scholarly discourse in the realm of AI's
influence on stock market.

2. Overview of the AI research

Artificial intelligence (AI) signifies a transformative shift in contemporary technology, embodying
the cognitive prowess exhibited by machineries, which is fundamentally different from the
intellectual capabilities of both humans and animals according to Nabeel (2023). The elucidation
of AI, as delineated in the Oxford English Dictionary, succinctly captures its essence as the
capability of machineries to perceive, amalgamate, and deduce information in a seamless manner.
Moreover, IBM elaborates on this notion by emphasizing AI's proficiency in replicating human
cognitive functions such as decision-making, troubleshooting, and linguistic analysis.
In the realm of financial services, AI has spring up as a formidable disruptive influence that is
fundamentally altering conventional paradigms and bringing about a revolutionary transformation
in well-established practices. Scholars exemplified by Han et al. (2023), Papanicolaou (2023), and
Mishra (2023) have meticulously chronicled the profound and far-reaching impact of AI on
decision-making processes, the enhancement of customer service offerings, and the methodologies
employed in assessing risks within the financial domain. The infusion of AI into the intricate
frameworks of finance has facilitated the development of customized and tailored strategies, the
provision of virtual interactions for clientele, and the detection of potentially fraudulent activities,
although this evolution is not without its share of challenges, including but not limited to cognitive
biases, vulnerabilities inherent in systems, and ethical considerations as discussed by Mahajan et
al. (2023). Furthermore, AI assumes a central and significant pivotal role in strengthening the
various protocols aimed at safeguarding cybersecurity within the financial sector, utilizing
sophisticated techniques such as data encryption and the utilization of machine learning algorithms
to effectively mitigate and respond to escalating cyber threats and to uphold the sanctity and
integrity of valuable data assets, as highlighted by Suresh et al. (2023).
A considerable amount of scholarly investigation has been dedicated to the examination of the
utilization of AI and ML techniques in the prediction of fluctuations in stock market volatility. ML
algorithms, especially sophisticated deep learning models such as ANN and LSTM networks, have
exhibited remarkable predictive capabilities when contrasted with traditional econometric models
(Sahiner et al., 2023; Yuyan et al., 2023; Yu et al., 2023). These AI-driven models excel in capturing
intricate interconnections and disparities present within stock market datasets, thereby amplifying
the accuracy of volatility forecasts (Lee et al., 2022; Seifi and Shavarani, 2023). Particularly
noteworthy is the outperformance of deep learning methodologies, notably the LSTM Neural
Network, over conventional models in the realm of predicting stock market volatility, thereby
emphasizing the efficacy of AI in the domain of asset management (Petrozziello et al., 2022). In
our research, we initiated a comprehensive search on the WOS. The selection of WOS was
motivated by its ability to reflect scholarly attention towards the effect of AI on the Stock Market.
When exploring the subject of "Impact of AI on Stock Market," we discovered a vast array of 3254
documents across various databases. After eliminating records that were not as significant, we
refined this number to 2055 articles for a more detailed examination. We gathered comprehensive
bibliographic data for these articles from WOS, encompassing details such as the title/ topic, author
& Contributors, citation, keywords, journal & publisher, abstract, issue year, and other publication
details. These evidence and records were then transferred to VOSviewer for a thorough analysis.
VOSviewer, a tool designed for analyzing scientific literature, aids in the visualization of trends
and patterns within the scientific literature. In this study, VOSviewer is employed to graphically
illustrate Complicated structures for statistical analysis and to perform cluster analysis.

[Link] search and selection

The initial phase of our research was initiated on March 25, 2023, with subsequent refinements
conducted on March 27, 2024, to identify pertinent studies. Our strategy involved utilizing the
WOS databases to encompass publications indexed within Research database. In our
comprehensive analysis, we employed a meticulously curated set of query keywords to ensure a
broad and nuanced exploration of the subject matter. This approach encompassed not only core
terms related to AI and full text Examination but also extended to include equivalents and
particular subjects pertinent to the field. The AI-related keywords included DL, ML, stock market,
risk management, portfolio optimization, investment optimization, prediction, and forecast. In our
comprehensive full text analysis, we employed a variety of techniques to dissect and understand
the textual data. These methods included full text Examination, Textual Examination, topic
modeling, NLP, word embedding, sentence embedding and sentiment analysis. To ensure the
accuracy and inclusivity of our search results, we strategically used asterisks (*) and quotation
marks (“”) in our queries. This approach was designed to account for plural forms, hyphens, or
spelling variations, thereby enhancing the robustness of our search criteria.
This methodical approach to keyword selection was designed to maximize the relevance and
comprehensiveness of our search results. By including a wide range of terms and their variations,
we aimed to capture a broad spectrum of information related to AI and text analysis, thereby
ensuring a comprehensive understanding of the subject matter. This strategy also facilitated the
identification of nuanced insights and trends within the field, enhancing the depth and breadth of
our analysis. On the other hand, inclusion of synonyms and specific topics was particularly crucial
in ensuring that our search was not limited to the most commonly used terms. By incorporating a
variety of terms and their variations, we were able to capture a broader range of information,
including less commonly discussed topics and nuanced aspects of the field. This comprehensive
approach allowed us to Gained a more nuanced understanding of the subject matter, enabling us
to identify and analyze trends and insights that might otherwise have been overlooked. Keywords
were searched in the title, abstract, keywords and keyword plus. The specific query is as follows.:
(“artificial intelligence*” OR “deep learning*” OR “machine learning*”) AND (stock market) .

Search query: ("artificial intelligence*" OR "deep learning*" OR " machine learning*") AND (stock market)
Identification

Research article records (n1) Other type of article records (n2)

Web of science core collection: Web of science core collection:

Non-English records n= 9 Title, abstract, author keywords and keywords plus
Title, abstract, author keywords and keywords Systematic review n2.1, Early access n2.2,
plus Proceeding Paper n2.3
n1= 1215 Correction, editorial material, Retracted n2.1= 50
Publication, data paper n= 14 n2.2= 95
n2.3= 7
Initial screening in citation manager based on Duplicates n= 356
duplicates, language, records type and availability Initial screening in citation manager based on
duplicates, language, records type and availability
n1= 846 n2.1= 47
Screening

Without full-text access n= 16

n2.2= 84
n2.3= 13

Title and abstract screening in database Title and abstract screening in database
Items unrelated to this research:
n2.1= 26
n1= 1420 n= 147 n2.2= 42
n2.3= 1

Full text assessment:

Eligibility

Full text assessment: Items unrelated to this research:

n= 38 n2.1= 21
n1= 1280
n2.2= 40
n2.3= 1

Publish records in Web of science core Included: Publish records in Web of science core Included:
n2.1= 20
n1= 1244 n2.2= 39
n2.3= 1
Included

Records Included in this systematic review:

n= 1304

Fig. 1 The flow diagram of the literature selection phases

The methodology employed for the literature search and selection process is detailed in Figure 1.
Initially, a comprehensive search was conducted within the WOS databases, yielding a total of
1304 records. The subsequent screening phase was meticulously designed to ensure the relevance
and quality of the selected articles. This phase involved the removal of articles based on several
criteria, including:
 Non-English papers/documents.
 Notes, editorials/ reports, conference proceedings titles, and preliminary papers/ documents.
 Duplicate/ equal entries.
 Documents without full-text view.
Furthermore, the topic/title and abstracts of the articles were scrutinized against Content-driven
exclusion criteria to eliminate documents that did not align with the study's focus on artificial
intelligence (AI) and text analysis. The decision to not set inclusion/exclusion criteria by discipline
was deliberate, aiming to capture a broad spectrum of perspectives on the subject matter. Articles
were only excluded if they failed to meet the following criteria:
 Insufficientdata concerning AI and text/full-text analysis.
 Focus solely on the technical aspects of AI.
 Lack of detail regarding the particular text analysis methods used.
Following the initial screening process, 1342 paper were identified for comprehensive full-text
evaluation. Upon reapplication of the exclusion criteria, a refined pool of n1=1244 and n2=60
article ultimately meeting the criteria for inclusion in the literature review. This meticulous process
ensures the rigorous selection of studies that not only align with the research question but also
contribute valuable insights to the field. The final selection of 1304 studies for the literature review
underscores the importance of adhering to strict inclusion and exclusion criteria to maintain the
integrity and relevance of the review. This approach not only Strengthen the quality of the literature
review but also Assists a more focused and impactful analysis of the existing body of knowledge.

3.1 The role of artificial intelligence in predicting volatility

AI's influence transcends merely predicting volatility to encompass the forecasting of market
trends by harnessing a wide array of extensive datasets sourced from various outlets such as news
articles, financial reports, and social networking platforms as discussed by Kanthimathi et al.
(2023). The utilization of deep learning methodologies, which encompass CNNs and recurrent
neural networks, plays a significant pivotal role in facilitating the extraction of complex dynamics
and correlations embedded within stock market data, a concept expounded upon by Nabiee and
Bagherzadeh (2023). Moreover, the execution of AI-powered ML models has demonstrated
remarkable precision in forecasting trends in Asian stock market volatility, thereby showcasing its
efficacy in dynamic relationship prognostication as highlighted in the research conducted by Lee
et al. (2022).
3.2 The role of AI in portfolio management & investment
In the realm of portfolio management, the utilization of AI-driven technologies has led to a
significant revolution in investment strategies, bringing about optimizations in risk-return
dynamics and enhancements in decision-making processes as evidenced by studies conducted by
Parisi and Manaog (2023) as well as Santos et al. (2022). The introduction of asset management
systems powered by AI has played a Significant pivotal role in instigating noteworthy
transformations in the field since the year 2008. These systems have been instrumental in
facilitating fundamental analysis and the development of strategies, all while navigating through
obstacles associated with model intricacies and the imperative need for transparency, a theme
explored by Bhuyan and Singh (2022). Moreover, the integration of genetic algorithms has further
elevated the process of investment portfolio development, providing invaluable guidance in asset
allocation strategies aimed at minimizing risks and maximizing returns, a phenomenon detailed in
the research by Adebiyi et al. (2022). Recent advancements in AI-driven portfolio optimization
models, including but not limited to PredACGAN and Dense Based EIIE, have been showcased
to outperform traditional strategies, a fact underscored by the empirical findings put forth by Kim
and Lee (2023). Furthermore, the deployment of sophisticated Deep Reinforcement Learning
(DRL) techniques in the context of portfolio optimization, exemplified by frameworks such as the
Dense Block Enhancement (DBE), has yielded substantial improvements in both cumulative return
metrics and Sharpe ratio measurements, a development explored by Gao et al. (2022).
In essence, the body of existing scholarly works emphasizes the profound and far-reaching
capacity of Artificial Intelligence (AI) within the realm of the financial industry, specifically in
terms of amplifying the efficacy of decision-making procedures and streamlining the management
of investment portfolios. Notwithstanding the obstacles encountered along the way, technologies
propelled by AI persist in revolutionizing the landscape of financial operations and methodologies,
thereby establishing a pathway for the emergence of novel, forward-thinking approaches and
fostering the sustenance of enduring economic advancement.

3.3 Overview of the current research of AI

Table 1 provides a detailed analysis of academic paper publication trends within the Web of
Science (WOS), detailing the total number of publications and specific counts across disciplines
including Computer Science Artificial Intelligence, Business Finance, Economics, and
Engineering Electrical Electronic. All articles extracted in WOS were published after 2015, when
the use of AI and its tools in the stock market was expanding. In the initial nine years of our
research, a considerable volume of papers were either made available online or cataloged by other
databases, which were not incorporated into our study. Our focus was exclusively on the Web of
Science (WOS), a leading high-level literature database, a methodology widely accepted in the
academic community for conducting literature reviews. This approach ensures the reliability and
relevance of the selected literature review (Ipek 2019). This data, compiled as of March 15, 2024,
indicates a notable decrease in publications for the year 2024 compared to previous years,
reflecting the ongoing expansion of the Artificial Intelligence field. The data collection period,
extending to March 27, 2024, accounts for the diminished article count in 2024, highlighting the
growth trajectory of AI research in the stock market. The continuous rise in publications since
2015 underscores the escalating interest and investment in AI research on stock market and its
pivotal role in both academic and professional discourse.
In a comprehensive analysis of 1890 articles, the prevalence of certain keywords within the
discourse on Artificial Intelligence (AI) was examined. The findings revealed that, alongside AI,
the terms "deep learning" (DL), "machine learning" (ML) and "stock market" were the most
frequently mentioned, occurring 421, 385, and 973 times, respectively. These results underscore
the central role of AI and its most prominent application within the stock market literature.

Table 1 Number of academic papers on Impact of artificial Intelligence on stock market

WOS- WOS -Computer Science WOS -Business WOS – WOS – Engineering
Articles Artificial Intelligence Finance Economics Electrical Electronic
2015 12 10 1 1 3
2016 22 9 1 3 3
2017 40 14 7 8 8
2018 50 22 5 4 13
2019 137 41 21 21 26
2020 222 56 22 32 36
2021 348 82 59 50 74
2022 478 102 71 63 78
2023 517 119 107 93 84
2024 64 11 14 17 5
Total 1890 455 308 292 330

3.4 Articles published in AI field

The main goal of our study was to explore the utilization of AI within the stock market domain.
To achieve this, in the present paper, we conducted a comprehensive review-analysis of scholarly
articles across various disciplines, including Computer Science Artificial Intelligence, Business
Finance, Economics and Engineering Electrical Electronic. Our search yielded a total of 455, 308,
292 and 330 articles from the Web of Science (WOS), respectively. This extensive collection of
literature allowed us to delve into the published journals, research topics, citations, and other
relevant metrics to provide a massive comprehensive understanding of the Present condition of AI
research in the stock market. Our analysis revealed a significant body of work that spans multiple
disciplines, highlighting the multifaceted nature of AI applications in the stock market. The
research topics covered a wide range of AI techniques and their implications for financial markets,
including stock price prediction, portfolio management, and risk assessment. The citations and
references within these articles further underscored the interdisciplinary nature of AI research in
finance, indicating a robust and evolving field of study.
The analysis also highlighted the importance of AI in enhancing the efficiency and accuracy of
stock market forecasting. For instance, AI-powered ETFs and AI stock pickers have emerged as
innovative solutions to traditional portfolio management and stock selection processes. These
advancements not only reduce costs and increase efficiency but also pave the way for more
sophisticated investment strategies that leverage big data and ML algorithms. In this research, we
have examined a variety of review articles focusing on the utilization of AI in the financial domain,
with a particular emphasis on those that delve into specific research topics rather than
encompassing a broad spectrum of subjects. For instance, Kumbure et al. (2022) conducted a
comprehensive literature review, focusing on the utilization of ML techniques for stock market
prediction, with a particular emphasis on the markets and variables utilized in these models.
Similarly, Li and Bastos (2020) conducted a systematic review on SM prediction using DL and
TA. These studies exemplify the depth and specificity with which AI applications in finance are
being explored, highlighting the growing interest and sophistication in this research area.

3.5 Research areas in AI field

Figure 2 illustrates the distribution of AI research across 25 primary domains. Notably, the three
primary areas of publication within AI research are distinctly highlighted through the use of
different colors. The leading domain is Computer Science Artificial Intelligence, with a total of
585 publications. This is followed by Electrical Electronic Engineering and Business Finance,
with 358 and 345 publications, respectively.

Fig. 2 The Types of research areas in AI field

3.6 Global analysis of countries-specific scholarly output in AI

This section delves into the analysis of scholarly output concerning the effect of AI on the stock
market, focusing on both small and big data perspectives. The preeminent contributors to this field,
as identified through our research, are China, the United States, India, South Korea, and England,
which rank as the top five countries in terms of published articles. Detailed statistics are
encapsulated in Table 2. Notably, the China has surpassed all other nations in terms of article
production, making up over one-third of the complete output. Finally, after collecting data and
information, Taiwan emerged as the sixth-ranked contributor with 103 papers. The comprehensive
dataset, encompassing 2414 articles, originates from 25 countries and regions.
In contrast, our exploration of the broader themes of "big data" and "artificial intelligence" yielded
a significantly larger volume of research. Specifically, we identified 2876 papers related to big
data, originating from 106 countries, alongside 1976 papers focusing on AI, DL and ML, sourced
from 46 countries. These findings underscore the nascent nature of AI research, highlighting the
need for increased international collaboration and scholarly engagement in this rapidly evolving
field.
 Table 2 Main research countries
Row Country No. of Papers %117.41
1 PEOPLES R CHINA 613 29.815%
2 USA 271 13.181%
3 INDIA 248 12.062%
4 SOUTH KOREA 139 6.761%
5 ENGLAND 126 6.128%
6 TAIWAN 103 5.010%
7 GERMANY 72 3.502%
8 AUSTRALIA 71 3.453%
9 ITALY 70 3.405%
10 SPAIN 64 3.113%
11 SAUDI ARABIA 63 3.064%
12 FRANCE 60 2.918%
13 IRAN 56 2.724%
14 PAKISTAN 52 2.529%
15 TURKEY 52 2.529%
16 CANADA 50 2.432%
17 BRAZIL 49 2.383%
18 JAPAN 45 2.189%
19 MALAYSIA 41 1.994%
20 SINGAPORE 36 1.751%
21 VIETNAM 36 1.751%
22 GREECE 27 1.313%
23 POLAND 27 1.313%
24 SOUTH AFRICA 22 1.070%
25 FINLAND 21 1.021%

3.7 Authors' published documents

This section provides an analysis of the documentary output by prominent authors, as depicted in
Figure 3. This visual representation meticulously tracks the temporal evolution of authorship, with
timelines rendered as lines, document production volumes represented by bubble sizes, and
citation intensity symbolized by color intensity. This comprehensive visualization not only
quantifies the volume of work produced by each author over time but also highlights the impact
of their contributions through the frequency of citations.
Among the contributors to this field, Wang J, Zhang Y, and Wang X stand out for their exceptional
publication rates. Their prolific output not only enriches the discourse but also sets a benchmark
for other scholars in the domain.
In 2017, the scholarly output by Wang J. was modest, with a single article and a total of five
citations annually. This trend saw a significant uptick in 2020, with three articles published and a
total of seventeen citations. The following year, 2021, marked a further increase, with five articles
and a total of 28.5 citations. The trajectory of growth continued into 2022, with ten articles
published and a total of 13.67 citations. The year 2023 witnessed an eleven-article contribution,
albeit with a decrease in citations to eight. Until March 2024, the tally stands at seven articles with
no citations recorded for the year. This data reflects a steady increase in the number of publications
by Wang J., accompanied by a fluctuating trend in citation rates, highlighting the evolving impact
and recognition of their work within the academic community.
In 2017, Zhang Y published one article, garnering a total of four citations. The following year, the
citation count increased to 7.33 with a single publication. The trend of growth continued in 2020,
with three articles published and a total of 24.4 citations. In 2021, we experienced a marginal
reduction in both the volume of articles and the number of citations, with two articles and 10.25
citations. The trajectory of growth resumed in 2022, with six articles published and a total of 30
citations. The year 2023 marked a significant milestone with 12 articles published, accumulating
a total of 37 citations. Until March 2024, Zhang Y has published two articles, yet to receive any
citations.
In the period from 2018 to 2024, the publication output and citation metrics for Wang X's work
exhibited a fluctuating trend. The initial year, 2018, saw a single article with a total of 16 citations.
The following year, 2019, marked a slight increase in both the number of articles (3) and the total
citations (38.33). The subsequent years, however, showed a more varied pattern. 2020 saw a
decrease in both metrics, with only 1 article and 5.6 citations. The trend reversed in 2021, with 3
articles and 8.25 citations. The year 2022 experienced a significant increase in both metrics, with
5 articles and 12 citations. The year 2023, however, saw a decrease in both metrics, with 4 articles
and 2 citations. Until March 2024, the publication count stands at 6 articles, but the citation count
remains at 0, indicating a potential shift in the academic reception of Wang X's work. This data
underscores the dynamic nature of academic citation trends, highlighting the importance of
ongoing research and publication efforts in maintaining and improving citation metrics.

Fig.3 Authors' published documents

3.8 Organizational affiliation overview
Figure 4 illustrates the affiliations with the highest number of published papers, notably
highlighting the Chinese Academy of Sciences as the leading contributor with 46 affiliations. As
anticipated, Chinese universities dominate the top 15 affiliations, showcasing a significant
presence in the field of AI research. This dominance underscores the substantial contribution of
Chinese institutions to the discourse on AI, particularly in the context of ML and DL-based data
analyses.
In Fact, the prominence of Chinese institutions in AI research is a testament to the country's
commitment to advancing technological innovation and fostering a robust academic environment.
This trend is indicative of the global shift towards multidisciplinary research, where AI
applications are increasingly being explored across various sectors, including engineering,
electronics, and business finance. The diversification of research areas within AI research reflects
a broader recognition of the multidisciplinary nature of AI technologies and their potential to
transform various industries.

Fig. 4 Documents by Organizational Affiliation

Furthermore, In the analysis of organizational dependency statistics, the Southwestern University

of Finance and Economics China and the University of Chinese Academy of Science (CAS) emerge
as significant contributors, securing the second and third positions with 25 and 22 dependency
statistics, respectively. This data underscores the pivotal role these institutions play in the field of
AI, particularly within the context of financial and economic research. The prominence of these
universities in AI research highlights the growing interdisciplinary nature of AI studies, extending
beyond traditional computer science domains to encompass finance, economics, and other related
fields. This trend reflects a broader shift in academic focus towards integrating AI technologies
into various sectors, emphasizing the multidisciplinary potential of AI research.

3.9 Most cited countries

Figure 5 illustrates the leading 15 nations with the most significant rates of citation, showcasing a
clear dominance by China, the United States, and India, which rank first, second, and third,
respectively, with 7031, 2810, and 2751 citations. This data underscores the significant influence
these nations exert in the academic and research community, highlighting their pivotal roles in
advancing knowledge and innovation across various disciplines.

Fig. 5 Top 15 most cited country.

3.10 Distribution of authors' keywords

Figure 6 illustrates a five-dimensional graph detailing the keyword contributions of authors from
2015 to March 2024. The analysis reveals three prevalent themes: ML, DL, and AI. These fields
have seen significant growth and application in the financial sector, underscoring their importance
in contemporary research and development.
In the current paper, we delve into the pivotal concepts of portfolio management, risk, volatility,
investment and forecasting, which have been consistently employed in the discourse of financial
literature over the years. These keywords underscore the enduring relevance of these topics in the
realm of financial risk management and investment strategies. The exploration of volatility
forecasting, in particular, highlights its critical role in financial risk management, as evidenced by
its frequent application in various studies and analyses. This reflects the ongoing interest and
necessity of understanding and managing volatility in financial markets, emphasizing the
importance of these concepts in shaping investment decisions and portfolio management
strategies.
 Fig. 6 The connection between keywords of authors over the time

3.11 Most cited articles

In the realm of AI, the integration of this emerging technology into various industries and fields
has garnered significant attention. This attention is reflected in the citation patterns of academic
papers, which serve as a barometer of influence and relevance in the research community. A
comprehensive analysis of 10 influential papers in the AI domain reveals a robust citation
landscape, underscoring the technology's pivotal role in shaping research across disciplines. These
10 papers collectively garnered 3,138 citations, with 729 of these citations excluding self-citations.
The papers themselves cited a total of 983 articles, of which 564 were non-self-citations. The most
cited articles, with more than 80 citations each, are detailed in Table 3, highlighting their
significant impact on the field. Notably, Fischer and Krauss (2018) emerged as the most popular
article within our dataset, boasting 845 citations in the WOS. This paper demonstrates the
effectiveness of LSTM networks in predicting out-of-sample directional volatilities & movements
for S&P 500 constituent stocks from 1992 to 2015, outperforming traditional memory-free
classification methods, and identifies high volatility and short-term reversal return profiles as key
predictors, suggesting a rules-based short-term reversal strategy for trading. This statistic
underscores the profound influence of their work on AI research, particularly in the context of
financial services, management, optimization, and prediction.
The exploration of AI's applications and implications across these domains is not only timely but
also crucial. As AI continues to evolve, its integration into existing frameworks and the
development of novel applications become increasingly pertinent. This analysis underscores the
necessity for continued exploration and dialogue within the research community, Striving to
leverage the capabilities of artificial intelligence to foster innovation and enhance productivity
across of stock market.
Table 3 The leading 10 scholarly paper with the highest aggregate citations and yearly citation counts
Author(S) Title Of Article Citation Summary

This study demonstrates the effectiveness of Long Short-Term Memory (LSTM) networks in predicting out-of-sample
Fischer and Deep learning with long short-term directional movements for S&P 500 constituent stocks from 1992 to 2015, outperforming traditional memory-free
Krauss memory networks for financial market 845 classification methods, and identifies high volatility and short-term reversal return profiles as key predictors, suggesting
(2018) predictions a rules-based short-term reversal strategy for trading.

This paper evaluates the effectiveness of four prediction models—Artificial Neural Network (ANN), Support Vector
Predicting stock and stock price index Machine (SVM), Random Forest, and Naive-Bayes—for predicting the direction of movement and stock price index for
Patel et al movement using Trend Deterministic 473 Indian stock markets, using two approaches for input data, demonstrating that Random Forest outperforms the other
(2015) Data Preparation and machine learning models when technical parameters are represented as continuous values, and all models improve performance when these
techniques parameters are represented as trend deterministic data.

Deep learning networks for stock market This study provides a comprehensive analysis of deep learning networks for stock market analysis and prediction,
Chong at al. analysis and prediction: Methodology, 351 highlighting their potential for high-frequency prediction by extracting features from raw data without prior knowledge
(2017) data representations, and case studies of predictors, and evaluating their performance through various data representation methods.

This paper presents a comprehensive review of computational intelligent methods applied in financial applications,
Cavalcante et Computational Intelligence and 314 focusing on studies published from 2009 to 2015, highlighting techniques for data preprocessing, clustering, forecasting,
al. (2016) Financial Markets: A Survey and Future and text mining, aiming to provide a systematic approach to building intelligent trading systems and discussing the field's
Directions main challenges and open problems.

This paper explores the prediction of future stock market index values using a two-stage fusion approach, incorporating
Patel et al. Predicting stock market index using 268 Support Vector Regression (SVR) in the first stage and combining it with Artificial Neural Network (ANN), Random
(2015) fusion of machine learning techniques Forest (RF), and SVR in the second stage, resulting in hybrid models for forecasting.

Deep neural networks, gradient-boosted This paper explores the effectiveness of deep neural networks, gradient-boosted trees, random forests, and their ensembles
Krauss et al. trees, random forests: Statistical 267 in statistical arbitrage, demonstrating promising empirical findings that challenge the semi-strong form of market
(2017) arbitrage on the S&P 500 efficiency, despite declining profits in recent years.

Despite significant challenges, the quest for models to predict financial market prices remains a highly researched area,
Henrique et Literature review: Machine learning with machine learning models, particularly support vector machines and neural networks, emerging as prevalent tools for
al. (2019) techniques applied to 204 recognizing complex patterns and predicting market values, especially within North American markets, highlighting the
financial market prediction ongoing relevance and potential of developing market data in this research domain.
Table 3 The top 10 most cited papers by total citation and citation per year

Author(S) Title Of Article Citation Summary

This paper introduces a novel end-to-end model, the Multi-Filters Neural Network (MFNN), specifically
Long et al. Deep learning-based feature designed for feature extraction in financial time series and price movement prediction, demonstrating superior
(2019) engineering for stock price movement 192 performance over traditional models in accuracy, profitability, and stability.
prediction

This paper aims to provide a comprehensive review of recent advancements in deep learning models for stock
market prediction, focusing on data sources, neural network structures, evaluation metrics, and the
Applications of deep learning in stock 136 implementation and reproducibility of these models, with the goal of facilitating synchronization with the latest
Jiang (2021)
market prediction: Recent progress progress and enabling easy reproduction of previous studies as baselines, while also highlighting future research
directions in this field.

This study contributes to the burgeoning literature on empirical asset pricing in the Chinese stock market by
constructing and analyzing a comprehensive set of return prediction factors using various machine learning
algorithms, highlighting liquidity as a crucial predictor and examining the impact of transaction costs, with a
Leippold et al. Machine learning in the Chinese stock 88 focus on the distinct characteristics of the Chinese market compared to the U.S. market, including the
(2022) market predictability of large stocks and state-owned enterprises over longer horizons.
3.12 Descriptive results of AI
This segment provides an exhaustive examination of the trends in the dissemination of research
within the domain of AI, covering a wide array of scholarly contributions such as scholarly articles,
sections of books, presentations at conferences, and preliminary studies. Figure 7 illustrates the
annual distribution of AI-related publications across various types of academic outputs and
research domains. This visual representation provides a clear overview of the growth trajectory
and diversification of AI research over the years.
600

500

400

300

200

100

0
2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
journal article book chapter conference paper working paper

Fig. 7 Distribution of different types of research

The initial phase of the study focuses on the earliest AI publications, beginning with the year 2015.
This period is particularly noteworthy as it coincides with the onset of a substantial increase in the
volume and diversity of AI-related research. The subsequent analysis aims to elucidate the
evolution of AI research trends, shedding light on the expanding landscape of AI studies and their
implications for the broader academic and professional communities.
The exponential growth in the annual publication of scholarly articles underscores the escalating
interest and acknowledgment of data analysis as a pivotal methodology within the realm of AI
research. Until 2018, conference proceedings served as the predominant source for disseminating
papers related to this field. In 2019, a significant transformation was observed, leading to an
increase in the publication of scholarly articles within academic journals. Over the years, papers
related to Computer Science artificial intelligence initially held the majority of the discussion. This
prominence can be credited to the rigorous coding skills required for numerous AI-based data
analysis tasks.
On the other hand, In the recent past, there has been a notable increase in scholarly contributions
from the fields of Engineering Electrical Electronics and Business Finance. Additionally, research
from disciplines such as Computer Science Information Systems and Economics has significantly
enriched this domain. This trend underscores a broadening diversification of research areas,
indicating a shift in interest from Computer Science Artificial Intelligence towards these emerging
fields.
3.13 AI publishers
In this section, we delve into the analysis of the publishers of scholarly papers on artificial
intelligence (AI), as depicted in Figure 8. Our findings reveal that the Elsevier database leads in
the publication of AI-related articles, with a total of 593 publications. Springer Nature follows
closely in second place, with 358 publications. Mdpi database in third place, with 202 publications.
These statistics underscore the distribution of AI research across various databases, highlighting
the most prominent platforms for the dissemination of AI-related scholarly work. The dominance
of the Elsevier database in AI publications is particularly noteworthy, given its significant role in
the dissemination of AI research. This database's prominence suggests a robust and growing
interest in AI research, as evidenced by the high volume of publications. On the other hand,
Springer Nature's and mdpi substantial presence in the AI research landscape further supports the
field's growing importance and the increasing number of researchers contributing to this area.
These findings are significant as they provide insights into the publishing landscape of AI research.
They indicate that while the Elsevier, Springer Nature and mdpi database are leading platforms for
AI-related publications, the field's research is distributed across a range of databases. This
distribution reflects the multidisciplinary nature of AI research and its relevance across various
sectors of the finance.
700

600

500

400

300

200

100

0
Elsevier Springer Mdpi IEEE Taylor & Wiley Hindawi Emerald World Public Library
Nature Francis Publishing Group Scientific Science
Group Publishing
Fig. 8 Distribution of different types of publishers in AI field

Moreover, the growing diversification of research areas within AI research, as evidenced by the
increasing number of publications in fields such as Electrical Engineering Electronics, and
Business Finance, underscores the evolving landscape of AI applications. This shift in research
focus is indicative of the growing recognition of AI's transformative potential across sectors, from
enhancing operational efficiency to driving economic diversification and fostering innovation.
In conclusion, the analysis of AI-related publications across various databases provides valuable
insights into the current condition of AI research. It highlights the importance of continued research
and development in AI technologies to enhance financial analysis and decision-making processes.
The findings also underscore the need for fostering interdisciplinary collaboration to fully realize
the potential of AI technologies.
3.14 Average citations per year
Figure 9 illustrates a significant trend in the academic discourse surrounding the integration of AI
within the stock market. The majority of references cited in the publications under study date back
to 2017 or later, with a notable peak in the citation rate for the year 2023, averaging approximately
550 citations. This data point underscores the growing interest and engagement in this area of
research. The oldest reference traced in these publications dates back to 2003, suggesting a gradual
but persistent exploration of the potential applications and implications of AI in financial markets.
This historical perspective provides a foundation for understanding the evolution of research in
this field, highlighting the increasing relevance and sophistication of AI-driven financial market
analyses and predictions.
600.00

500.00

400.00

300.00

200.00

100.00

0.00
1992 1993 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024

Fig. 9 Average Citations Per Year

3.15 Analyzed the network of papers’ keywords

This section presents an in-depth analysis of the keyword network within a collection of scholarly
papers, as depicted in Figure 10. The visualization of this network is designed to elucidate the
frequency of keywords, their co-occurrence within papers, and the temporal distribution of their
appearance. The size of each node in the network corresponds to the frequency of a keyword, while
the connections between nodes indicate the co-occurrence of keywords within a single paper.
Additionally, the color of each node is used to represent the average year in which the keyword
appears, providing a temporal context to the keyword usage.
The analysis reveals that the three core concepts of AI: ML, DL and AI itself are the most frequently
mentioned keywords across the papers. This finding underscores the centrality of these AI concepts
within the discourse on their applications in the stock market. The prominence of these keywords
suggests a robust and growing interest in the exploration and development of AI technologies,
particularly in the context of stock markets. The keyword network analysis not only highlights the
prevalence of AI-related concepts but also offers insights into the interconnectedness of these
concepts within the research landscape. The co-occurrence of keywords indicates a shared focus
on integrating AI technologies into financial analysis and prediction, reflecting a collective effort
to leverage AI's capabilities to enhance market understanding and decision-making processes.
 Fig. 10 The network of analysis of keywords co-
occurrence

Moreover, the temporal distribution of keywords, as indicated by the color coding of nodes,
suggests a dynamic and evolving field of study. The increasing frequency of AI-related keywords
over time indicates a growing recognition of the potential of AI in transforming financial markets.
This trend is likely driven by advancements in AI technologies, the accumulation of data, and the
increasing sophistication of financial models that incorporate AI components. The analysis of the
keyword network thus provides a comprehensive overview of the current condition of AI research
within the stock market. It highlights the importance of ML, DL, and AI as key concepts in this
field, underscoring the need for continued research and development in these areas. The findings
also suggest a promising future for the integration of AI in financial markets, where AI
technologies are poised to play a very Significant pivotal role in shaping market trends and
informing investment decisions. The analysis reveals that machine learning is the earliest concept
to appear in the dataset, with its applications spanning various aspects of the stock market,
including returns, information, news, Twitter, sentiment analysis, finance, portfolio management,
and prediction. This early emergence of machine learning as a keyword in the context of the stock
market underscores its foundational role in the field. The co-occurrence of machine learning with
other keywords suggests a multifaceted approach to leveraging AI technologies in financial
analysis and decision-making processes.
Deep learning, identified as the second most frequently occurring concept in our analysis, is
prominently associated with domains such as risk, volatility, and media, highlighting its pivotal
role in these areas. This association underscores the significant impact of deep learning on financial
risk control, market analysis, and quantitative investment, as well as its potential in predicting
market volatility and integrating with financial time series for intelligent investment and risk
management. The integration of DL into stock market not only enhances the accuracy of market
volatility prediction but also significantly reduces human costs and improves economic hazard
control and business processing capabilities. This integration marks a new era of innovation and
change in the financial territory, where DL serves as a catalyst for more intelligent management
and production methods. Furthermore, the keyword "artificial intelligence" is frequently
associated with specific applications in finance and markets, topic modeling, and relationship
analysis techniques such as ensemble methods and reinforcement analysis.
The analysis of keyword associations within the realm of artificial intelligence (AI) reveals distinct
patterns that correlate with specific financial activities, analyses, and approaches. While specific
terms related to full-text analysis are infrequently highlighted, they are consistently present across
various AI scopes. Sentiment analysis, in particular, is closely associated with DL and ML,
indicating a synergy between these techniques in financial contexts. Conversely, topic modeling
and its associated keywords are more closely linked to broader AI applications, highlighting the
diversity and interconnectedness of AI research within the financial domain.

3.16 Top publishers in AI

This section presents a Full-scale analysis of the publication landscape concerning AI and its
applications in the stock market, Concentrating on the distribution of articles across various
journals. Our study identified top 10 publisher that have published papers on this topic, with Figure
8 providing a detailed list of these publishers and the count of articles issued in each. Table 4
further highlights the most influential journals in the field of AI research, including prominent
publishers such as Elsevier, Springer Nature, MDPI, Taylor & Francis, Wiley, and IEEE. As we
move beyond 2015, there has been a notable increase in the publication of articles related to AI in
the stock market. The period from 2020 to 2023 has seen the topmost amount of articles on this
subject, underscoring the growing significance of AI and its sub-disciplines, including deep
learning and machine learning, in the financial sector. This trend reflects the evolving role of AI
in enhancing market analysis and prediction, as well as its potential to revolutionize financial
forecasting and disclosure practices. The rise in AI-related publications is indicative of the field's
rapid development and the increasing recognition of its importance in the stock market. This
growth is supported by advancements in ML and DL, which enables the analysis of vast datasets
to detect meaningful patterns and trends, thereby offering a more nuanced understanding of market
performance. Predictive analytics, a component of ML and DL, uses historical data to forecast
future financial trends, enabling informed decisions on revenue projections, market share, and
risks.
The burgeoning interest in Artificial Intelligence (AI) and its applications within the stock market
is not only evidenced by the proliferation of published articles but also by the escalating frequency
of AI mentions during earnings calls. This trend, which has seen a notable increase from
approximately 478 mentions in 2022 to 516 mentions by 2023, signifies a pivotal shift from
conventional financial metrics towards the adoption of more sophisticated AI-driven analytics.
This development underscores the revolutionary capabilities of AI within the financial sector,
marking a significant stride towards the integration of advanced technologies in financial decision-
making processes.
Table 4 Top journals publishing research
Elsevier No. of Papers Springer Nature No. of Papers

EXPERT SYSTEMS WITH APPLICATIONS 95 COMPUTATIONAL ECONOMICS 19

APPLIED SOFT COMPUTING 26 NEURAL COMPUTING & APPLICATIONS 17

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 25 SOFT COMPUTING 16

KNOWLEDGE-BASED SYSTEMS 13 APPLIED INTELLIGENCE 13

FINANCE RESEARCH LETTERS 12 MULTIMEDIA TOOLS AND APPLICATIONS 12

INTERNATIONAL REVIEW OF FINANCIAL ANALYSIS 11 FINANCIAL INNOVATION 10

INFORMATION SCIENCES 10 ANNALS OF OPERATIONS RESEARCH 7

NEUROCOMPUTING 8 JOURNAL OF BIG DATA 6

INFORMATION PROCESSING & MANAGEMENT 6 COGNITIVE COMPUTATION 5

PATTERN RECOGNITION 5 JOURNAL OF SUPERCOMPUTING 4

ENERGY ECONOMICS 4 ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING 3

MDPI No. of Papers Taylor & Francis No. of Papers

MATHEMATICS 21 APPLIED ARTIFICIAL INTELLIGENCE 8

APPLIED SCIENCES-BASEL 17 QUANTITATIVE FINANCE 7

ELECTRONICS 11 APPLIED ECONOMICS LETTERS 5

JOURNAL OF RISK AND FINANCIAL MANAGEMENT 4 EMERGING MARKETS FINANCE AND TRADE 3

AXIOMS 3 JOURNAL OF EXPERIMENTAL & THEORETICAL ARTIFICIAL 2

INTELLIGENCE

ECONOMETRICS 2 ECONOMIC RESEARCH-EKONOMSKA ISTRAZIVANJA 2

Wiley No. of Papers IEEE No. of Papers

JOURNAL OF FORECASTING 8 IEEE ACCESS 51

CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE 5 IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL 4

SYSTEMS

INTELLIGENT SYSTEMS IN ACCOUNTING FINANCE & MANAGEMENT 4 IEEE TRANSACTIONS ON NEURAL NETWORKS AND 3
LEARNING SYSTEMS
3.17 Thematic evolution of AI
In our study, we constructed a thematic evolution map (Figure 11) to elucidate the burgeoning
relevance and escalating scholarly interest in specific research domains. Our analysis revealed a
pronounced shift towards AI, ML, and DL analytics, with these fields receiving the majority of the
research attention. This trend underscores the growing of AI, ML, and DL in the scientific
community, highlighting their pivotal role in contemporary research endeavors. The upper right
quadrant of our analysis primarily concentrates on the utilization of ML in stock market
forecasting, encompassing a range of topics including volatility, asset allocation, and the
exploration of deep reinforcement learning methodologies. This focus underscores the
transformative potential of machine learning in enhancing financial market predictions, while also
acknowledging the broader scope of financial modeling and strategy development.

Fig. 11 AI research themes between 2022-2024 (February).

In the lower right quadrant, the emphasis shifts towards the exploration of NLP and AI, particularly
through the Long Short-Term Memory (LSTM) framework. This area also encompasses, albeit to
a lesser extent, discussions on deep reinforcement learning. These advancements in AI and ML are
pivotal in the financial sector, offering novel approaches to data analysis and prediction, thereby
enhancing decision-making processes and strategic planning. In the lower left quadrant, the
discussion encompasses the domains of production efficiency, portfolio management, trading
strategies, and deep reinforcement learning. These areas represent a critical intersection of
operational efficiency, financial asset management, algorithmic trading, and advanced machine
learning techniques. Production efficiency is a cornerstone of operational excellence, ensuring
resources are utilized optimally. Portfolio management, a critical component of financial planning,
involves the strategic allocation of assets to achieve desired returns and risk levels. Trading
strategies, particularly those employing algorithmic approaches, leverage data and analytics to
make informed decisions in the market. Deep reinforcement learning, a cutting-edge machine
learning technique, is increasingly applied to enhance these strategies by learning from data to
make more informed and potentially profitable trading decisions. This quadrant underscores the
convergence of technology and finance, highlighting the potential for innovative solutions in asset
management and trading.
In the upper left quadrant, the discussion encompasses the integration of ML models within asset
management, alongside the exploration of risk management and the analysis of network
complexity factors. This section delves into the application of advanced algorithms to optimize
asset allocation, while simultaneously addressing the multifaceted nature of risk and the intricate
dynamics of network complexity.

3.18 Factorial analysis

Figure 12 illustrates the standard and prevalent conceptual frameworks derived from a
comprehensive analysis of 49 keyword terms, categorized into seven distinct clusters through the
application of Factorial Analysis and Multiple Correspondence Analysis (MCA). This
methodological approach facilitates the identification of underlying patterns and relationships
within the data, thereby enhancing the interpretability and utility of the retrieved keyword. In the
analysis, Bibliometrix was utilized to assign cluster colors, adhering to the default color palette
sequence of red, blue, green, purple, orange, and so forth. This methodology ensures a consistent
and visually appealing representation of the data clusters. The domain of management, particularly
within the context of the current research, encompasses a broad spectrum of disciplines, including
portfolio management, and volatility analysis. This area has also been delineated in the realm of
market forecasting, underscoring its significance in the financial landscape. The intricate
relationship between these elements is crucial for investors, financial analysts, and policymakers
alike, as it provides a framework for understanding and navigating the complexities of stock
market.

Fig. 12 Factorial analysis using management analysis method for 49 keywords distributed in 7 clusters
Volatility forecasting, a critical component of this domain, has evolved significantly over the years,
with advancements in estimation methods and data accessibility playing a pivotal role. The
importance of capturing both intraday and overnight market movements for effective risk
management cannot be overstated. Sophisticated forecasting techniques, which place a greater
emphasis on recent observations, have demonstrated superior performance compared to traditional
methods. Moreover, incorporating high-frequency return data has been shown to significantly
enhance the accuracy of volatility forecasts, thereby offering a more nuanced understanding of
market risk. Investors and financial professionals are increasingly relying on volatility forecasting
to inform their investment strategies and risk management practices. By accurately estimating
market volatility, investors can better prepare for potential market fluctuations and make more
informed decisions regarding portfolio diversification and risk tolerance. This process is not
without its challenges, as the measurement of forecast accuracy and the interpretation of volatility
data require a deep understanding of complex models and methodologies.
in conclusion, the field of management, particularly in the areas of portfolio management and
volatility analysis, is thus intertwined with market forecasting. This interplay is essential for the
development of robust investment strategies and the effective management of financial risks. As
financial markets continue to evolve, the importance of these disciplines will only grow,
underscoring the need for ongoing research and innovation in this domain.

3.19 Collaboration network between institutions

The institutional collaborative network map, as depicted in Figure 13, showcases a significant
cluster of collaboration among prominent academic institutions. This cluster is particularly
noteworthy due to its substantial presence in the WOS database. The primary focus of this
collaboration is the Southwestern University of Finance and Economics in China, alongside the
University of Chinese Academy of Sciences and Xiamen University. The collaboration network
map, a visualization tool derived from bibliometric analysis, is instrumental in identifying the key
contributors and institutions within a specific research domain. By sizing nodes based on
publication or citation counts and edges based on the frequency of collaborations, this map
provides a clear representation of the central authors or institutions within the network. The
presence of clusters in the map not only highlights the research communities that have converged
around specific topics but also reveals the interconnections among these communities. Moreover,
the absence of edges between adjacent nodes or clusters may indicate potential avenues for future
collaboration, thereby offering insights into the dynamics of academic collaboration within the
field of finance and economics.
The significance of this collaboration network map lies in its ability to not only identify the most
influential institutions within the field of AI studies in finance but also to provide a visual
representation of the research community's structure. This visualization tool is crucial for academic
institutions seeking to demonstrate the value of their research, especially in challenging financial
times. on the other hand, the significance of collaboration in the realm of AI within financial
markets, particularly the stock market, cannot be overstated. This collaboration is not only evident
among authors hailing from the same country but belonging to different institutions, but it also
transcends national borders, encompassing researchers and scientists from various countries and
institutions. This global interconnectivity underscores the internationalization of AI applications
in financial markets, highlighting the technology's pervasive influence on global financial systems.

Fig. 13 Institutions collaboration networks

The interplay between these two facets—the global collaboration in AI application within financial
markets and the educational efforts to bridge the gap in AI adoption—illustrates a dynamic and
evolving landscape in the financial sector. As AI continues to transform financial markets, the
importance of fostering international collaboration and addressing educational challenges becomes
increasingly critical. This dual focus ensures that the financial industry can harness the full
potential of AI, while also preparing future generations of professionals to navigate this rapidly
evolving technological landscape.

3.20 Author collaboration networks

Co-authorship network analysis serves as a critical tool for elucidating the interconnectedness
among researchers, as evidenced by the volume of publications they have jointly authored. This
methodology, rooted in the retrieval and standardization of scientific publications, allows for the
visualization and quantification of these networks, thereby facilitating a comprehensive
understanding of the collaborative landscape within the academic community. By examining the
structure and dynamics of these networks, researchers can uncover the underlying mechanisms
that shape the scientific community, including the patterns of collaboration and the distribution of
research outputs. This analysis not only highlights the significance of co-authorship as a proxy for
research collaboration but also underscores its potential to reveal the broader implications of
scientific collaboration on the production and dissemination of knowledge. As the field of co-
authorship network analysis continues to evolve, it opens new avenues for exploration, offering
comprehension and insights into the complexities of research collaboration and its effect on the
advancement of scientific disciplines.
The analysis of influential authors, as depicted in Figure 14, unveils a multifaceted network of
connections that transcends the boundaries of the initially identified clusters. This intricate
network, derived from the Web of Science (WOS) data, highlights the expansive influence of
certain authors beyond their immediate peer group. A prime example is the extensive network of
Wang J., which extends beyond his cluster, encompassing collaborations with authors from various
other clusters. The WOS-derived map is segmented into smaller clusters, each representing distinct
groups of authors that do not interconnect. This fragmentation suggests a pattern of isolated
collaboration, where authors operate independently within their specific clusters.
This complex web of connections underscores the dynamic nature of academic collaboration,
where authors may find themselves bridging gaps between different research communities. The
fragmentation into smaller clusters could be indicative of a broader trend towards specialization,
where authors focus on niche areas of research, potentially leading to more isolated but impactful
contributions. The review and analysis carried out in the current research provides valuable
insights into the evolving landscape of academic collaboration, highlighting the importance of
interdisciplinary efforts in advancing knowledge.

Fig. 14 Author collaboration networks

3.21 Collaboration world map

In the realm of AI research, the significance of international collaboration is underscored by the
fact that over 20% of the published articles feature co-authorship from multiple countries. This
trend highlights the critical role of international co-authorship in advancing scholarly contributions
within the AI field. Figure 15 elucidates the intricate network of international collaborations
among nations. The most dynamic regions for global partnerships are highlighted as China, the
United States, Europe, Australia, South Africa, Japan, Korea, Canada, Vietnam and Singapore.
These nations are at the forefront of international cooperation, underscoring their significant roles
in shaping global alliances and initiatives.

Fig. 15 Collaboration world map

Artificial Intelligence (AI) emerges as a pivotal field in the realm of economic development, with
its applications significantly influencing a nation's economic trajectory. This burgeoning domain
of research has seen a predominant focus on collaborative endeavors among developed nations.
The interdisciplinary nature of AI's impact on economic activities necessitates a comprehensive
understanding of its current status, potential, and future research directions. This understanding is
crucial for identifying knowledge gaps and formulating strategies for future research. The
collaboration between scholars from different regions, particularly between the USA and China,
underscores the global significance of AI in economic development. This collaboration not only
strengthens the intellectual landscape of AI and economic development but also fosters a vibrant
exchange of ideas and methodologies across borders.

4. Discussion
The aim of a systematic review is to carefully identify, assess, and combine the results of all
previous reviews on a specific subject. This type of study is crucial because it provides a broad
overview of the current knowledge on the topic. By combining and analyzing the results from all
relevant analysis and reviews, we can gain a complete understanding of the field. This method and
approach helps to clarify the current knowledge and pinpoint areas that need further investigation.
Additionally, a systematic review can identify areas where studies agree or disagree. It helps in
resolving conflicting findings and highlights areas where more research is needed to address
inconsistencies. The systematic review as a critical tool in advancing knowledge within a specific
domain. It not only identifies gaps in the research but also underscores areas where consensus
exists. This dual role is crucial for guiding future research efforts, ensuring that the field progresses
in a coherent and meaningful manner. By synthesizing the findings of numerous reviews, this
approach allows for a nuanced understanding of the current landscape of knowledge. It enables
researchers to discern patterns, trends, and areas of contention within the literature, thereby
informing future research directions. This analysis acts as a crucial instrument for steering future
studies and investigation by identifying specific regions that need further research, exploration and
suggesting possible avenues for future studies. It ensures that future research is solidly grounded
and tackles essential questions and investigations that have not been thoroughly examined.
Furthermore, by giving a thorough summary of the present knowledge landscape and highlighting
research voids, this analysis elevates the caliber of future research. It assures that future studies are
solidly grounded and tackle very important questions and uncertainties that have not been
adequately addressed. Such a methodology, in turn, aids in advancing understanding in the realm
of stock market forecasting through AI (Lin & Marques, 2024).
In essence, the systematic reviews is a powerful tool for synthesizing and interpreting the vast
body of literature on a given topic. It provides a massive comprehensive review of the current
condition of knowledge, identifies gaps and areas of consensus, and highlights areas where
additional research is required. This approach not only aids in the clarification of the existing body
of knowledge but also serves as a roadmap for future research endeavors. Prior systematic reviews
have compiled statistical findings related to the utilization of AI in stock price prediction and
market analysis. However, these reviews often present fragmented insights into the complexities
of AI applications in stock prediction performance, risk management, portfolio optimization, and
investment strategies. This research endeavors to conduct a thorough examination and synthesis
of systematic reviews to pinpoint the most efficient AI techniques, data references, and
performance metrics for advancing stock market prediction, management, and optimization. This
exhaustive analysis is vital for guiding future strategies in the stock market, highlighting the pivotal
role of AI in financial investment. The research questions that guide this study are meticulously
crafted to identify the most widely used AI methods, informational resources, and performance
indicators within the domain of stock market forecasting, management, and optimization. By
addressing these questions, this study seeks to contribute to the ongoing discourse on the role of
AI in financial markets, offering valuable insights that could inform the development of more
sophisticated and effective investment optimization. The significance of this research lies in its
potential to bridge the gap between theoretical discussions on AI applications in finance and
practical implementations in stock market strategies. By identifying the most effective AI
approaches and performance metrics, this study aims to provide a foundation for future research
and innovation in the field of financial investment. This systematic review, confined to
publications between 2015 and March 2024, reveals that research focusing on the utilization of AI
for stock market forecasting analysis is relatively recent. Despite the inclusion and exclusion
criteria applied, the remaining articles, spanning from 2020 to 2023, underscore the novelty of this
research domain. A noteworthy observation is the prevalence of studies on AI methodologies in
stock market research, despite the absence of AI technology descriptions in the search terms. As
illustrated in Figure 7, there has been a notable increase in publications over time.
In this segment, we aim to explore the following inquiries:
  [Link] are the prevalent AI methodologies and technologies used for forecasting stock
market trends?
 [Link] AI methodologies and technologies are most frequently applied for predicting
stock market volatility?
 [Link] AI methodologies and technologies are commonly used for risk management within
the stock market?
 [Link] AI methodologies and technologies are most commonly applied for portfolio
optimization in the stock market?
 [Link] AI methodologies and technologies are most commonly utilized for investment
optimization in the stock market?
We foresee that our deliberations will significantly contribute to the advancement of academic
research and the development of practical applications in the future.
RQ1 - What are the prevalent AI methodologies and technologies used for forecasting stock
market trends?
The most commonly utilized AI methods and technologies for predicting stock market trends
include ML algorithms, including LSTM models, and GNNs. These techniques have been
extensively employed due to their ability to handle the complex, nonlinear, and time-series nature
of stock market data. Machine Learning, particularly LSTM models, has emerged as a powerful
tool for predicting stock prices and trends. LSTM models are particularly adept at capturing the
temporal dependencies in stock market data, making them highly effective for time-series
forecasting. This method has been applied in various studies to predict stock prices with high
accuracy, demonstrating its effectiveness in capturing the intricate patterns within stock market
data. GNNs, including GCN and GAN, have also been utilized for stock market prediction. These
models are particularly suited for analyzing the complex relationships and dependencies within
financial data, making them a valuable tool for predicting stock market trends. GCN and GAT have
been identified as the most frequently utilized GNNs in stock market prediction, highlighting their
effectiveness in handling the complex, graph-structured data of financial markets. In addition to
LSTM and GNNs, other AI techniques such as SVM, ANN, and DRL have also been employed
for stock market prediction. These methods offer various advantages, including the ability to
handle high-dimensional data, the capacity for feature extraction, and the potential for dynamic
adjustment of investment strategies.
RQ2 - Which AI methodologies and technologies are most frequently applied for predicting
stock market volatility?
The most commonly utilized AI methods and technologies for predicting stock market volatilities
include DL, NLP, RL, SA, and QC. These advancements have significantly improved the accuracy
and data-driven nature of investment decisions, making AI an indispensable tool for navigating the
complexities of the stock market with precision and confidence. High-frequency trading (HFT) is
another critical application of AI in stock market prediction, where AI algorithms execute trades
within milliseconds, capitalizing on minuscule price discrepancies. This approach enhances the
ability to analyze market data and execute trades at high speeds with better accuracy. AI models
also play a crucial role in portfolio management by analyzing historical market data and volatility,
adjusting portfolios in real-time to align with changing market conditions, and suggesting
diversification strategies to mitigate potential risk. Technical analysis is another area where AI has
made significant contributions. AI-driven algorithms can analyze technical indicators such as
EMA, RSI, Bollinger bands, Fibonacci retracement, stochastic oscillator, and average directional
index to make accurate predictions about future price movements. Furthermore, AI's impartiality,
free from cognitive biases, human emotions, and other psychological factors, provides an objective
perspective, benefiting investors, traders, and financial institutions with objective and rational
insights, resulting in optimal investment decisions. ML methods, a branch of AI, offer a promising
solution for accurately forecasting stock prices by aggregating various factors to improve stock
return predictions. These methods have been shown to significantly outperform traditional
methods, achieving remarkable accuracy in predicting stock returns.
RQ3 - What AI methodologies and technologies are commonly used for risk management
within the stock market?
In the realm of stock market risk management, a variety of AI techniques and technologies are
frequently employed, such as ML and DL algorithms, NLP, predictive analytics, and data
visualization tools. These technologies facilitate the real-time analysis of extensive financial data,
enabling the detection of patterns, trends, and potential risks. Machine learning algorithms, in
particular, excel in analyzing historical data to predict future price fluctuations and market
volatility, thereby improving risk management strategies. DL algorithms, leveraging complex
neural networks, are proficient in deriving valuable insights from unstructured data types,
including trends, text, full text, audio, diagrams, figures, graphics and images, offering insights
into market dynamics and investor sentiment. NLP empowers computers to understand human
languages in news, documents, events, shocks and other forms of data, aiding in the identification
of events that can influence market movements. Predictive analytics and, utilizing ML, DL, data
mining, and statistical modeling, are crucial for forecasting future price movements and market
volatility based on historical data and current trends. Data visualization tools assist trading
professionals in comprehending intricate data sets and learning from AI-generated predicts and
recommendations. Moreover, the combination of AI algorithms with risk management systems
allows for the monitoring of trading activities and the evaluation of potential risks, thereby
enhancing market predictability, stability, and efficiency.
RQ4 - Which AI methodologies and technologies are most commonly applied for portfolio
optimization/management in the stock market?
In the domain of portfolio optimization, several sophisticated methodologies have gained
prominence, including Random Forest, SVM, and NN. Additionally, Reinforcement Learning,
NLP, and Sentiment Analysis offer unique approaches to this field. These methodologies are adept
at scrutinizing extensive datasets to uncover patterns and associations that can guide investment
decisions. The effectiveness of these methodologies in portfolio optimization underscores their
capacity to navigate intricate data and deliver practical insights. Random Forest (RF) stands out as
a versatile machine learning technique that is particularly adept at managing large datasets with
high dimensionality. It is particularly skilled at pinpointing patterns and associations within data,
making it an optimal choice for portfolio optimization. The operation of Random Forest algorithms
involves the creation of multiple decision trees and the consolidation of their outcomes to produce
a final prediction. This ensemble technique ensures reliable predictions, even when the data
contains noise or outliers. SVM represent another category of ML algorithms that have found
utility in portfolio optimization. SVMs are supervised learning models utilized for both
classification and regression analysis. They function by locating the hyperplane that most
effectively segregates the data into categories. In the context of portfolio optimization, SVMs can
be employed to categorize assets based on their performance attributes, thereby facilitating the
selection of the most suitable investments. Neural Networks (NN), a subset of machine learning
algorithms, draw inspiration from the structure and function of the human brain. They comprise
interconnected nodes, or "neurons," that process information. Neural Networks are particularly
proficient at identifying complex patterns and associations within data, making them highly
effective in portfolio optimization. They can learn from historical data to forecast future trends,
enabling investors to make informed decisions regarding asset allocation. RL is a form of ML
where an agent learns to make decisions by interacting with its environment. In the context of
portfolio optimization, reinforcement learning algorithms can adapt to fluctuating market
conditions and adjust the portfolio accordingly.
Techniques in Natural Language Processing (NLP) are employed to scrutinize textual information
such as news stories, financial reports, and social media opinions. By deriving significant insights
from unstructured information, NLP aids investors in making well-informed decisions. Sentiment
analysis (SA) encompasses evaluating public feelings towards stocks or the market as a whole. By
scrutinizing social media posts, news stories, and other sources, sentiment analysis offers crucial
data for portfolio optimization. The utilization of these ML methods in portfolio optimization
highlights their capability to augment investment decision-making processes. By harnessing the
capabilities of these algorithms, investors can obtain insights into market trends, pinpoint potential
investment opportunities, and manage risk more efficiently. This not only elevates the efficiency
of investment strategies but also enhances the overall performance of portfolios.

RQ5 - What AI methodologies and technologies are most commonly utilized for investment
optimization in the stock market?
The most commonly utilized AI methods and technologies for investment optimization in the stock
market include:
Predictive Analytics (PA): This tool uses ML algorithms to scrutinize historical market data,
aiming to discern patterns that could foreshadow future price fluctuations. This methodology offers
invaluable insights to investors by utilizing statistical modeling, data mining techniques, and
machine learning to anticipate future market movements based on past trends and data patterns.
Sentiment Analysis (SA): This technique evaluates the emotional tone of news articles, social
media posts, and other online content to determine public sentiment towards specific stocks or
markets. It serves as an additional criterion in the investment decision-making process, providing
insights into market sentiment that can influence investment strategies.
Quantitative Analysis (QA): Involves using mathematical models and statistical techniques to
evaluate securities, helping investors make decisions based on solid statistical evidence.

5. Conclusion
This paper presents a comprehensive systematic review examining the influence of AI on the stock
market, analyzing 2,053 articles related to AI within the WOS database. The selection of articles
was conducted from the WOS platform, focusing on documents with the highest citation counts to
identify the most pertinent research. The articles were subsequently categorized into four main
thematic areas: portfolio optimization, stock market prediction, risk management, optimization
investment using AI. In fact, the systematic review methodology ensures a rigorous and unbiased
analysis of the literature. By focusing on the most cited articles, the review aims to highlight the
most influential research in the field. The categorization of articles into specific thematic areas
allows for a focused examination of the impact of AI on different aspects of the stock market. The
analysis provides a comprehensive overview of the scholarly landscape concerning AI,
encompassing the most frequently cited papers across all time and those published annually. This
data serves as a critical indicator of the evolving interest in AI as a subject of investment, a trend
that parallels the technological advancements and widespread adoption of computing technologies.
The exponential growth in the number of papers published from 2015 to March 2024 underscores
AI's emergence as a rapidly expanding field of research, reflecting its increasing relevance and
appeal to scholars and practitioners alike. Given the emergence of artificial intelligence (AI) as a
pivotal technology, its application across various sectors within the stock market is of paramount
importance. Consequently, the exploration of AI's commercial applications is deemed crucial for
both academic and practical endeavors, underscoring the significance of this research area in the
broader context of financial investment and market analysis. For each area of study, we analyzed
the most cited papers from 2015 to 2024. Additionally, recent articles were identified to highlight
the latest findings and suggest potential future research directions. Tables were constructed to
visually distinguish the various methodologies and models presented in each paper, facilitating a
clear understanding of the research landscape.
We identify several promising research directions. The first significant area of focus is Risk
Management, where AI plays a crucial role in enhancing risk management practices. By utilizing
advanced analytics, AI can evaluate risk tolerance and suggest diversification strategies that align
with individual investor profiles. Machine learning models, trained on historical data, are capable
of analyzing patterns to identify potential risks, thereby facilitating proactive adjustments to
portfolios to mitigate these risks. This approach not only improves risk management but also
optimizes risk-adjusted returns, positioning AI as a proactive guardian in the investment landscape.
The second section pertains to the optimization of portfolios and investments. 2.1. Investment
Optimization: AI equips investors with sophisticated decision-making tools by leveraging machine
learning models to scrutinize extensive data for investment opportunities, risk assessment, and
portfolio optimization. NLP extracts insights from textual data, such as news and filings, offering
a comprehensive market perspective. Predictive analytics aim to forecast future market trends,
emphasizing explainability to enhance user trust. AI's capacity to analyze historical data, market
trends, and correlations enables the identification and evaluation of factors such as value, size,
momentum, quality, or volatility, facilitating the construction of targeted portfolios. [Link]
Optimization: The application of AI in portfolio optimization involves the analysis of extensive
datasets of historical asset performance and risk characteristics, including volatility and
correlation. This examination aids in the construction of tailored portfolios, focusing on low-risk
investments such as index funds and government bonds, in accordance with the principles of MPT.
By incorporating Risk Parity Optimization with artificial intelligence (AI) models, capital
allocation is determined by the risk contribution rather than the monetary value, providing a more
balanced risk profile for cautious investors.
The third potential research direction explores the prediction of stock market volatilities and
trends. 3.1. Stock Market Trend Prediction: The utilization of AI in predicting stock market trends
employs ML algorithms to analyze historical data, market trends, and correlations. This analysis
identifies and assesses factors such as value, size, momentum, quality, and volatility, enabling
targeted portfolio construction and the forecasting of future market trends. AI's predictive analytics
focus on ensuring explainability to build user trust. AI-powered factor investing models have
demonstrated significant performance improvements over traditional models, offering substantial
benefits to investors. 3.2. Forecasting Stock Market Volatilities: AI algorithms utilize historical
records, news, papers, social media opinions, and economic signals to Recognize patterns and
forecast potential market volatility. LSTM networks, a type of RNN , are employed for their ability
to learn long-term dependencies within market data, potentially leading to more accurate market
predictions. XAI techniques are being developed to enhance transparency in AI-driven market
forecasting, thereby fostering trust among investors.
This study acknowledges several limitations that are essential to consider in the context of its
findings and implications. Firstly, the analysis is confined to the literature available in the WOS
databases. This limitation may result in an incomplete representation of the relevant literature,
potentially overlooking significant studies that are not indexed within this database. Secondly, the
selection of literature was guided by subject categories within the WOS. This approach, while
methodologically sound, may have inadvertently excluded pertinent research that does not neatly
fit into the predefined categories. This limitation underscores the potential for bias in the literature
review process, which could affect the comprehensiveness and relevance of the findings. Lastly,
the recommendations derived from this study are inherently subjective. They are based on the
authors' interpretations and analyses of the existing literature, which are influenced by their
perspectives and the limitations of the data available. Recognizing these subjective limitations, the
authors express a commitment to fostering further research and scholarly discourse on these topics.
This commitment is aimed at enhancing the understanding and addressing the points raised in this
study. In this manner, it aids in enhancing understanding within the domain of AI and the stock
market.
In conclusion, while this study provides valuable insights into the subject matter, it is essential to
recognize and acknowledge its limitations. These limitations underscore the importance of
ongoing research and critical engagement with the literature to refine our understanding and to
address the gaps identified in this study.
Abbreviations
WOS Web Of Science
AI Artificial intelligence
DL Deep Learning
ML Machine Learning
XAI Explainable AI
LSTM Long Short-Term Memory
RNN Recurrent Neural Network
NLP Natural Language Processing
MPT Modern Portfolio Theory
SA Sentiment Analysis
QA Quantitative Analysis
PA Predictive Analytics
SVM Support Vector Machines
EMA exponential moving average
RSI relative strength index
HFT High-frequency trading
DRL Deep Reinforcement Learning
ANN Artificial Neural Networks
NN Neural Networks
RL Reinforcement learning
RF Random Forest
GNN Graph Neural Networks
GAN Graph Attention Networks
GCN Graph Convolutional Networks
CNN Convolutional Neural Networks
SM Stock Market
TA Technical Analysis
GAN Graph Attention Networks
ANN Artificial Neural Networks
SVM Support Vector Machines
DRL Deep reinforcement learning
QC Quantum Computing

Declaration
Availability of data and materials
Data used in this paper were collected from Web of Science Core Collection

Competing interests
The authors declare that they have no competing interests.

Funding
Not applicable.
Acknowledgements
We are grateful to the 4th international conference on management, business, economic and
accounting for their review and acceptance of our paper.

References
Lin, C. Y., & Marques, J. A. L. (2024). Stock market prediction using artificial intelligence: A systematic review of systematic
reviews. Social Sciences & Humanities Open, 9, 100864. [Link]
Najem, R., Amr, M. F., Bahnasse, A., & Talea, M. (2024). Advancements in Artificial Intelligence and Machine Learning for Stock
Market Prediction: A Comprehensive Analysis of Techniques and Case Studies. Procedia Computer Science, 231, 198-204.
[Link]
Lin, C. Y., & Marques, J. A. L. (2024). Stock market prediction using artificial intelligence: A systematic review of systematic
reviews. Social Sciences & Humanities Open, 9, 100864. Available at SSRN: [Link] or
[Link]
Jareño, F., & Yousaf, I. (2023). Artificial intelligence-based tokens: Fresh evidence of connectedness with artificial intelligence-
based equities. International Review of Financial Analysis, 89, 102826. [Link]
Nabeel, R. (2023). Artificial intelligence (AI) is intelligence—perceiving, synthesizing, and inferring information—demonstrated
by machines, as opposed to intelligence displayed by non-human animals and humans. Example tasks in which this is done
include speech recognition, computer vision, translation between (natural) languages, as well as other mappings of inputs.
The Oxford English Dictionary of Oxford University Press defines artificial intelligence as (No. sz7mj). Center for Open
Science. doi: 10.31219/[Link]/sz7mj.
Han, Y., Chen, J., Dou, M., Wang, J., & Feng, K. (2023). The Impact of Artificial Intelligence on the Financial Services Industry.
Academic Journal of Management and Social Sciences, 2(3), 83-85. doi: [Link]
Papanicolaou, P. (2023). How AI is re-shaping the financial industry: lights, shades and perspectives. In Artificial Intelligence and
Financial Behaviour (pp. 205-214). Edward Elgar Publishing. [Link]
Mishra, S. (2023). Exploring the impact of ai-based cyber security financial sector management. Applied Sciences, 13(10), 5875.
[Link]
Mahajan, D., Vatsyayan, s., Kumar, s., Dadhich, p. (2023). Decision Strategies and Artificial Intelligence Navigating the Business
Landscape. doi:10.59646/edbook/009.
Abakah, E. J. A., Tiwari, A. K., Ghosh, S., & Doğan, B. (2023). Dynamic effect of Bitcoin, fintech and artificial intelligence stocks
on eco-friendly assets, Islamic stocks and conventional financial markets: Another look using quantile-based approaches.
Technological Forecasting and Social Change, 192, 122566. doi: 10.1016/[Link].2023.122566.
Suresh, N., Neelam, H., Chakrapani, E., Kumar, K. A., & Ali, S. S. (2023, January). Artificial Intelligence Advances and Their
Repercussions on the Financial System. In 2023 International Conference on Computer Communication and Informatics
(ICCCI) (pp. 1-4). IEEE.
Jang, J., & Seong, N. (2023). Deep reinforcement learning for stock portfolio optimization by connecting with modern portfolio
theory. Expert Systems with Applications, 218, 119556. [Link]
Sahiner, M., McMillan, D. G., & Kambouroudis, D. (2023). Do artificial neural networks provide improved volatility forecasts:
Evidence from Asian markets. Journal of Economics and Finance, 47(3), 723-762. doi: 10.1007/s12197-023-09629-8.
Yuyan, G., Di, H., Yan, M., & Hongmin, Z. (2023). Realised volatility prediction of high-frequency data with jumps based on
machine learning. Connection Science, 35(1), 2210265. [Link]
Yu, Y., Lin, Y., Hou, X., & Zhang, X. (2023). Novel optimization approach for realized volatility forecast of stock price index based
on deep reinforcement learning model. Expert Systems with Applications, 233, 120880.
[Link]
Seifi, N., & Shavarani, H. S. (2023). Predicting Idiosyncratic Volatility from Stock Market Trade Records: A Machine Learning
Approach. [Link]
Kanthimathi, N., Saranya, N., & Hermina, C. I. (2023, May). Stock Market Prediction Using AI. In 2023 2nd International
Conference on Vision Towards Emerging Trends in Communication and Networking Technologies (ViTECoN) (pp. 1-5).
IEEE. doi:10.1109/ViTECoN58111.2023.10157327.
Nabiee, S., & Bagherzadeh, N. (2023). Stock Trend Prediction: A Semantic Segmentation Approach. arXiv preprint
arXiv:2303.09323.
[Link]
Mita, D., & Takahashi, A. (2023). Multi-agent model based proactive risk management for equity investment. Engineering
Applications of Artificial Intelligence, 125, 106701. [Link]
Parisi, L., & Manaog, M. L. (2023). Optimal Machine Learning-and Deep Learning-driven algorithms for predicting the future
value of investments: A systematic review and meta-analysis. doi:10.21203/[Link]-2658566/v1.
Kim, J., & Lee, M. (2023). Portfolio optimization using predictive auxiliary classifier generative adversarial networks. Engineering
Applications of Artificial Intelligence, 125, 106739. [Link]
Gao, R., Gu, F., Sun, R., Stefanidis, A., Ren, X., & Su, J. (2022, October). A Novel DenseNet-based Deep Reinforcement
Framework for Portfolio Management. In 2022 International Conference on Cyber-Enabled Distributed Computing and
Knowledge Discovery (CyberC) (pp. 158-165). IEEE.
Bhuyan, B. P., & Singh, T. P. (2022). Artificial intelligence in financial portfolio management. In Revolutionizing Business
Practices Through Artificial Intelligence and Data-Rich Environments (pp. 108-139). IGI Global. doi: 10.4018/978-1-6684-
4950-9.ch007.
Adebiyi, S. O., Ogunbiyi, O. O., & Amole, B. B. (2022). Artificial intelligence model for building investment portfolio optimization
mix using historical stock prices data. Rajagiri Management Journal, 16(1), 36-62. doi:10.1108/RAMJ-07-2020-0036.
 Santos, G. C., Barboza, F., Veiga, A. C. P., & Gomes, K. (2022). Portfolio optimization using artificial intelligence: a systematic
literature review. Exacta. doi:10.5585/exactaep.2022.21882.
Petrozziello, A., Troiano, L., Serra, A., Jordanov, I., Storti, G., Tagliaferri, R., & La Rocca, M. (2022). Deep learning for volatility
forecasting in asset management. Soft Computing, 26(17), 8553-8574. doi:10.1007/s00500-022-07161-1.
Song, G. H., & Jain, A. (2022). Can artificial intelligence beat the stock market?. Studies in Economics and Finance, 39(5), 772-
785.
doi: 10.1108/SEF-03-2022-0133.
Lee, P., Huang, Z., & Tang, Y. (2022). Trend Prediction Model of Asian Stock Market Volatility Dynamic Relationship Based on
Machine Learning. Security and Communication Networks, 2022. [Link]
Chaajer, P., Shah, M., & Kshirsagar, A. (2022). The applications of artificial neural networks, support vector machines, and long–
short term memory for stock market prediction. Decision Analytics Journal, 2, 100015.
[Link]
Gonzales, R. M. D., & Hargreaves, C. A. (2022). How can we use artificial intelligence for stock recommendation and risk
management? A proposed decision support system. International Journal of Information Management Data Insights, 2(2),
100130. doi:10.1016/[Link].2022.100130.
Rubesam, A. (2022). Machine learning portfolios with equal risk contributions: Evidence from the Brazilian market. Emerging
Markets Review, 51, 100891. [Link]
Leippold, M., Wang, Q., & Zhou, W. (2022). Machine learning in the Chinese stock market. Journal of Financial Economics,
145(2), 64-82. [Link]
Jiang, W. (2021). Applications of deep learning in stock market prediction: recent progress. Expert Systems with Applications, 184,
115537. [Link]
Samitas, A., Kampouris, E., & Kenourgios, D. (2020). Machine learning as an early warning system to predict financial crisis.
International Review of Financial Analysis, 71, 101507. [Link]
Henrique, B. M., Sobreiro, V. A., & Kimura, H. (2019). Literature review: Machine learning techniques applied to financial market
prediction. Expert Systems with Applications, 124, 226-251. [Link]
Long, W., Lu, Z., & Cui, L. (2019). Deep learning-based feature engineering for stock price movement prediction. Knowledge-
Based Systems, 164, 163-173. [Link]
İpek, İ. (2019). Organizational learning in exporting: A bibliometric analysis and critical review of the empirical research.
International Business Review, 28(3), 544-559. [Link]
Fischer, T., & Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European
journal of operational research, 270(2), 654-669. [Link]
Chong, E., Han, C., & Park, F. C. (2017). Deep learning networks for stock market analysis and prediction: Methodology, data
representations, and case studies. Expert Systems with Applications, 83, 187-205.
[Link]
Krauss, C., Do, X. A., & Huck, N. (2017). Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on
the S&P 500. European Journal of Operational Research, 259(2), 689-702. [Link]
Cavalcante, R. C., Brasileiro, R. C., Souza, V. L., Nobrega, J. P., & Oliveira, A. L. (2016). Computational intelligence and financial
markets: A survey and future directions. Expert Systems with Applications, 55, 194-211.
[Link]
Patel, J., Shah, S., Thakkar, P., & Kotecha, K. (2015). Predicting stock and stock price index movement using trend deterministic
data preparation and machine learning techniques. Expert systems with applications, 42(1), 259-268.
[Link]
Patel, J., Shah, S., Thakkar, P., & Kotecha, K. (2015). Predicting stock market index using fusion of machine learning techniques.
Expert systems with applications, 42(4), 2162-2172. [Link]

View publication stats