INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue XI November 2025

Artificial Intelligence in Stock Market Trading - A Comprehensive

Survey of Models

Cynthia Udoka Duruemeruo, Adetunji Oludele Adebayo, Nathaniel Akande; Olatunde Olasehan; Uju

Judith Eziokwu

Department of Computer Science, University of Wolverhampton, United Kingdom

DOI: https://dx.doi.org/10.51244/IJRSI.2025.12110019

Received: 18 November 2025; Accepted: 27 November 2025; Published: 03 December 2025

ABSTRACT

With financial markets becoming increasingly complex and volatile, traditional statistical and human-driven

approaches are proving inadequate for capturing the nonlinear and dynamic nature of stock trading. Artificial

Intelligence (AI) has therefore emerged as a transformative force, employing advanced algorithms and deep

learning to identify hidden patterns, forecast prices, and inform trading decisions. This paper presents a

comprehensive bibliometric survey of 9,088 scholarly works spanning 1971 to 2025, offering the most detailed

review of AI applications in stock market trading to date. Using the SPAR-4-SLR framework and bibliometric

tools such as Biblioshiny and VOS viewer, the study maps intellectual contributions, identifies key research

clusters, and analyses collaboration networks across the field. The results reveal dominant methodologies

including neural networks, long short-term memory (LSTM), reinforcement learning, and hybrid approaches,

while also highlighting the growing importance of Explainable AI (XAI) and ESG-aligned frameworks.

Contributions from East Asian institutions, particularly in China, stand out, although significant inputs from

Europe and North America are also observed. Despite these advances, challenges persist in areas such as

interpretability, real-time adaptability, and the integration of alternative data sources like sentiment analysis and

satellite imagery. Future research directions emphasize the development of quantum AI, reinforcement learning-

based adaptive systems, and ethical regulatory frameworks that ensure responsible innovation. By bridging

theory and practice, this study provides an intellectual roadmap and practical recommendations for researchers,

practitioners, and policymakers. Overall, the findings underscore the urgency of advancing transparency,

robustness, and interdisciplinary collaboration to ensure AI-driven trading systems contribute to sustainable

financial innovation and trustworthy decision-making.

Keywords: artificial intelligence; stock market trading; machine learning; deep learning (LSTM/CNN);

reinforcement learning; explainable AI (XAI); bibliometrics; algorithmic trading

With financial markets growing more complex and volatile, traditional statistical and human-driven approaches

struggle to capture the nonlinear and dynamic nature of stock trading. Artificial Intelligence (AI) has emerged

as a transformative force, employing advanced algorithms and deep learning to identify hidden patterns, forecast

prices, and inform decision-making. This paper conducts a comprehensive bibliometric survey of 9,088 scholarly

works spanning 1971 to 2025, offering the most extensive review of AI applications in stock market trading to

date. Using the SPAR-4-SLR framework and bibliometric tools such as Biblioshiny and VOSviewer, the study

maps intellectual contributions, identifies key research clusters, and analyses collaboration networks. Results

reveal dominant methodologies including neural networks, long short-term memory (LSTM), reinforcement

learning, and hybrid approaches, while also highlighting the growing importance of Explainable AI (XAI) and

ESG-aligned models. East Asian institutions, especially in China, are leading global contributions, with

significant input from Europe and North America. Challenges remain around interpretability, real-time

adaptability, and integration of alternative data such as sentiment and satellite imagery. Future directions

emphasize quantum AI, reinforcement learning, and ethical regulatory frameworks. By bridging theory and

practice, this study provides both an intellectual roadmap and practical recommendations for researchers,

industry practitioners, and policymakers. The findings underscore the urgency of advancing transparency,

robustness, and interdisciplinary collaboration to ensure AI-driven trading systems contribute responsibly to

sustainable financial innovation.

Page 205

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue XI November 2025

INTRODUCTION

With the ever-changing financial markets becoming more and more complex and volatile, stock market trading

and prediction now become an arduous task for potential investors, consultants, and academicians. Traditional

methods of market analysis relying on tried and tested statistical formulas and human analogy cannot easily

capture the nonlinear and noisy structure of financial data. In this season, AI and deep learning have become

game-changing technologies, employing sophisticated algorithms to analyze enormous datasets for patterns and

make signal-based price forecasts. Despite its promise, the adaptation of AI for stock trading remains a problem,

for example in explaining models, affecting hypermarkets, and handling different data sources, including

contrasts in social media sentiments and satellite images in this field.

This research paper addresses the critical need for a systematic review of AI techniques in stock trading. The

study maps the intellectual landscape of the field by analyzing 9,088 documents from 1971 to 2025, disclosing

its very rapid development at 10.6% annual increases and a very high academic impact of 15.62 average citations

per document. It reveals dominant research clusters, including market applications, stock-specific predictive

modeling, and AI/ML methodologies, illuminating their interactions between theoretical advancements and real-

world applications. Key methodologies, including neural networks (NN), long-short-term memory (LSTM), and

hybrid models, are surveyed together with emerging trends such as transformer architectures and Explainable

AI (XAI).

The study also underscores the global nature of research in this domain, with strong contributions from East

Asian institutions like Beihang University and influential authors such as WANG J and LI Y. Challenges such

as "unexplainable" AI and market volatility are discussed, with proposed solutions including hybrid approaches,

real-time adaptive models, and ethical AI frameworks aligned with ESG (Environmental, Social, Governance)

goals. By bridging gaps between theory and practice, this paper not only consolidates decades of research but

also provides a roadmap for future innovation, emphasizing the need for transparency, robustness, and

interdisciplinary collaboration in AI-driven financial decision-making.

Through bibliometric analysis and thematic mapping, the research demonstrates how AI is reshaping stock

trading, from foundational algorithms to cutting-edge applications, while calling for a balanced approach that

prioritizes both technological innovation and ethical accountability in the financial sector.

LITERATURE REVIEW

Bibliometrics has emerged as a powerful quantitative methodology for analyzing scholarly publications, closely

related to fields like "infometrics" (Egghe & Rousseau, 1990; Wolfram, 2003) and "scientometrics" [4]. This

approach systematically examines various forms of academic output, including journal articles, books, patents,

dissertations, and grey literature, while its counterpart "webometrics" extends this analysis to digital content.

Originally focused on basic bibliographic metrics like author productivity and citation counts, bibliometric

studies have evolved to encompass geographical distributions, institutional contributions, and discipline-specific

developments (Lin, 2012; Zhuang et al., 2013; Huffman et al., 2013; Liu et al., 2012). [3] [16]

Modern bibliometric research leverages sophisticated tools such as Scopus, Gephi [5], and VOSviewer, enabling

comprehensive analyses of citation networks, co-authorship patterns, and thematic trends. These methods now

incorporate alternative metrics like download statistics and social media engagement alongside traditional

citation analysis [17]. However, researchers must exercise caution in data normalisation (Pellegrino, 2011) [15]

to ensure valid cross-disciplinary comparisons, given the methodology's reliance on large datasets. [5]

While bibliometrics has proven valuable for assessing research impact, institutional performance, and academic

productivity through citation analysis [15], it has faced criticism for potential over-reliance on quantitative

metrics. The Leiden Manifesto [10] notably cautions against allowing numerical data to overshadow qualitative

scholarly judgment. Despite these concerns, bibliometric techniques have gained prominence in business

research, facilitated by analytical tools like VOSviewer, Leximancer, and SciVal that can process extensive

publication databases. [10]

Page 206

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Contemporary applications of bibliometrics include tracking emerging research trends, analyzing collaboration

networks, and mapping the intellectual structure of academic fields. The methodology's effectiveness, however,

can be limited by incomplete data or methodological constraints. For comprehensive insights, researchers should

combine performance analysis with science mapping techniques to provide a more holistic understanding of

research landscapes.

RESEARCH METHODOLOGY

This study employs a systematic bibliometric analysis to map the intellectual landscape of AI applications in

stock market trading, aligning with the SPAR4SLR framework (Scientific Procedures and Rationales for

Systematic Literature Review) (Alonso-Garcia et al., 2021; Paul et al., 2021; Kunisch et al., 2023). The

methodology is structured into three phases: data collection, data organization, and bibliometric analysis.

[1][14][12]

In the data collection phase, Scopus was selected as the primary database due to its extensive coverage of high-

impact journals and rigorous indexing standards (Fahimnia et al., 2015; Fan et al., 2022), ensuring access to

seminal works in AI, finance, and computational economics. A Boolean search query ("artificial intelligence"

OR "machine learning" OR "deep learning" OR "neural networks" OR "reinforcement learning") AND ("stock

market" OR "stock trading" OR "financial markets" OR "algorithmic trading" OR "quantitative trading") AND

("models" OR "applications" OR "survey" OR "review") was designed to capture three thematic clusters: (1)

AI/ML techniques ("artificial intelligence" OR "machine learning" OR "deep learning" OR "neural networks"

OR "reinforcement learning"), (2) financial context ("stock market" OR "stock trading" OR "financial markets"

OR "algorithmic trading" OR "quantitative trading"), and (3) research focus ("models" OR "applications" OR

"survey" OR "review"). The initial search yielded 9,088 articles, which were manually screened to exclude

irrelevant fields (e.g., medicine) and retain only those directly related to AI-driven stock trading models. [8][9]

For data organisation, strict inclusion and exclusion criteria were applied. Only peer-reviewed journal articles

and reviews in English were considered, with a focus on AI/ML applications in stock trading, such as predictive

modelling, risk assessment, and high-frequency trading. Non-English publications, conference abstracts, and

financial studies not centred on AI were excluded to maintain thematic relevance.

The final phase involved bibliometric analysis, conducted using Biblioshiny in R Studio [2]. This tool generated

visualizations, including thematic maps, co-word networks, and citation trajectories, to uncover key trends and

collaboration patterns. The analysis identified prominent author and institutional networks, revealing influential

research clusters and emerging trajectories in AI-driven stock market trading. This structured approach ensures

a comprehensive and replicable review of the field’s intellectual structure.

RESULTS AND DISCUSSION

Table 1. Main Information

Page 207

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

The timespan of the data (Table 1), ranging from 1971 to 2025, indicates a long-standing interest in the

intersection of artificial intelligence (AI) and stock market trading, with a significant acceleration in research

output in recent years. The dataset comprises 9,088 documents sourced from 3,164 journals, books, and other

publications, reflecting a broad and diverse academic engagement with the topic. The annual growth rate of

10.6% underscores the increasing relevance and rapid development of AI applications in stock market trading,

driven by advancements in machine learning, deep learning, and computational power.

The average age of documents (5.34 years) suggests that the field is relatively dynamic, with a focus on recent

research. This is further supported by the high average number of citations per document (15.62), indicating that

the work in this area is highly influential and frequently referenced. The substantial number of references

(227,702) highlights the interdisciplinary nature of the field, drawing from computer science, finance,

economics, and statistics. The analysis of document contents reveals a rich keyword landscape, with 19,818

Keywords Plus (ID) and 13,123 Author's Keywords (DE), reflecting the diversity of research themes and

applications within AI-driven stock market trading.

The authorship analysis reveals a highly collaborative research environment, with 16,393 authors contributing

to the field. While single-authored documents account for 924 entries, the majority of research is collaborative,

with an average of 3.05 co-authors per document. International co-authorships constitute 15.26% of the total,

indicating a moderate level of global collaboration. This international dimension is crucial for the cross-

pollination of ideas and methodologies, particularly in a field as complex and multifaceted as AI in stock market

trading.

The document types are varied, with conference papers (4,322) and articles (3,893) dominating the landscape,

reflecting the field's emphasis on presenting new findings and models at academic conferences and in peer-

reviewed journals. The presence of books (66), book chapters (302), and reviews (125) indicates a growing effort

to consolidate knowledge and provide comprehensive overviews of the field. The inclusion of retracted

documents (24) and errata (16) highlights the rigorous scrutiny and self-correcting nature of the academic

process in this domain.

Figure 1. Annual Scientific Production.

Overall, the bibliometric analysis (Figure 1) paints a picture of a vibrant and rapidly evolving field characterised

by high levels of collaboration, interdisciplinary engagement, and a strong focus on recent advancements. The

findings underscore the importance of continued research and innovation in AI applications for stock market

trading, with significant potential for future growth and impact.

Page 208

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 2. Three Field Map: Cited Sources - Abstract - Country.

The cited sources (Figure 2) indicate a strong foundation in established financial theories and models, with

references to influential works by authors such as Fama, Engle, and others. This suggests that the research builds

on well-regarded studies in finance and economics, integrating them with advanced AI methodologies.

The abstract in the middle highlights the primary focus areas of the research, including stock market prediction,

financial modeling, and machine learning applications. Keywords such as "stock," "china," "model," "market,"

"data," "learning," and "prediction" dominate, indicating a significant emphasis on using AI techniques to

analyze and predict market behaviors. The presence of terms like "neural network," "forecasting," and "machine"

underscores the reliance on sophisticated computational models to enhance trading strategies and financial

decision-making.

On the right, the list of countries reflects a global interest in AI applications within stock market trading. The

inclusion of countries like the United Kingdom, Korea, Iran, Malaysia, Japan, Brazil, Indonesia, Australia, Saudi

Arabia, Canada, Bangladesh, Thailand, and Germany indicates widespread research activity and collaboration

across diverse geographical regions. This international dimension suggests that the challenges and opportunities

presented by AI in stock market trading are being explored from multiple cultural and economic perspectives,

enriching the field with a variety of insights and approaches.

Overall, the field plot illustrates a robust and interdisciplinary research landscape, where traditional financial

theories are being augmented with cutting-edge AI technologies. The global collaboration and diverse research

focus areas highlight the dynamic and evolving nature of the field, pointing towards continued growth and

innovation in the application of AI to stock market trading.

Figure 3. Most Relevant Sources.

Page 209

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

The bibliometric analysis highlights the most relevant sources (Figure 3) that have significantly contributed to

the field. The prominent sources include Lecture Notes in Computer Science, Expert Systems with Applications,

and IEEE Access, which are among the top venues for publishing cutting-edge research in AI and computational

finance. These sources are known for their rigorous peer-review processes and high impact, indicating that the

research in this domain is both credible and influential.

The presence of conference series such as ACM International Conference Proceeding Series and Lecture Notes

in Networks and Systems underscores the importance of academic conferences in disseminating new findings

and fostering collaborations. These platforms often serve as early venues for presenting innovative models and

applications, reflecting the rapid pace of advancements in AI-driven stock market trading. Additionally, journals

like Neural Computing and Applications and Neurocomputing emphasize the role of neural networks and

machine learning techniques in financial forecasting and decision-making.

The inclusion of multidisciplinary sources such as Mathematics and Applied Soft Computing indicates the

integration of mathematical models and computational techniques in addressing complex financial problems.

This interdisciplinary approach is crucial for developing robust AI models that can adapt to the dynamic nature

of stock markets. The variety of sources, ranging from specialized journals to broad-scope conference

proceedings, highlights the diverse methodologies and applications being explored in the field.

Overall, the most relevant sources reflect a vibrant and interdisciplinary research ecosystem, where traditional

financial theories are being enhanced by advanced AI technologies. The prominence of high-impact journals and

conferences underscores the field's rapid growth and the significant interest from the academic and professional

communities in leveraging AI for stock market trading.

Figure 4. Most Relevant Authors.

The bibliometric analysis identifies the most relevant authors (Figure 4) who have made significant contributions

to the field. Among the prominent authors are WANG J, LI Y, WANG Y, and ZHANG Y, who appear frequently

in the dataset, indicating their substantial influence and productivity in this domain.

The recurrence of surnames like WANG, LI, and ZHANG suggests a strong presence of researchers from East

Asia, particularly China, which aligns with the global trend of significant contributions from this region in AI

and computational finance. Their expertise likely spans various aspects of AI, including the development of

advanced algorithms for market prediction, optimization of trading strategies, and the application of deep

learning techniques to financial data.

Authors such as KUMARA and SINGH S indicate contributions from South Asia, highlighting the global nature

of research in this field. Their work might focus on integrating traditional financial theories with modern AI

techniques, contributing to a more comprehensive understanding of market dynamics.

Page 210

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

The high number of documents attributed to these authors (ranging from 75 to 100) underscores their active

engagement and leadership in advancing the field. Their research likely covers a broad spectrum of topics, from

theoretical models to practical applications, providing valuable insights and methodologies that drive innovation

in AI-driven stock market trading.

Overall, the most relevant authors represent a diverse and highly skilled group of researchers whose work is

pivotal in shaping the future of AI applications in financial markets. Their contributions not only advance

academic knowledge but also have practical implications for developing more accurate and efficient trading

systems.

Figure 5. Most Relevant Affiliations.

The bibliometric analysis highlights the most relevant affiliations (Figure 5) contributing to the field. Leading

institutions such as Beihang University, Peking University, and Tsinghua University are prominently featured,

indicating their significant role in advancing research on AI applications in stock market trading. These

universities are known for their strong emphasis on engineering, computer science, and financial technologies,

which aligns with the interdisciplinary nature of the research.

The presence of Southwestern University of Finance and Economics and Central University of Finance and

Economics underscores the importance of specialized financial institutions in this domain. These universities

likely contribute expertise in financial modeling, economic theory, and market analysis, complementing the

technical advancements in AI and machine learning.

International contributions from institutions like Waseda University in Japan and Islamic Azad University in

Iran reflect the global collaboration and diverse perspectives in the field. These affiliations highlight the

widespread interest in applying AI to financial markets across different economic and cultural contexts.

The inclusion of National Institute of Technology and Nanyang Technological University further emphasizes

the role of technological and engineering-focused institutions in developing innovative AI-driven trading

models. Their expertise in computational methods and data analysis is crucial for creating robust and efficient

systems for market prediction and trading.

Overall, the most relevant affiliations represent a mix of top-tier universities, specialized financial institutions,

and international collaborators, all contributing to the rapid advancement of AI in stock market trading. Their

collective efforts drive both theoretical innovations and practical applications, shaping the future of financial

technologies.

Page 211

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 6. Most Global Cited Documents.

The most globally cited documents (Figure 6) in the bibliometric analysis highlight key trends and influential

works in the application of Artificial Intelligence (AI) in stock market trading. Notably, foundational works like

CAMPBELL, JY (2012) on econometrics and HOCHREITER S (1997) on neural computation underscore the

integration of traditional financial theories with advanced AI techniques. The presence of multiple papers from

venues like Expert Systems with Applications and Neurocomputing emphasizes the dominance of machine

learning and expert systems in this domain.

Recent works, such as FISCHER T (2018) and PATEL J (2015), reflect the growing use of AI for predictive

analytics and decision support in trading. Meanwhile, early contributions like TAY FEH (2001) and KIM K-J

(2003) demonstrate the long-standing focus on optimization and neural networks. The diversity of models—

from game theory (CESA-BIANCHI N, 2006) to deep learning (NELSON DMO, 2017)—suggests a

multidisciplinary approach, combining econometrics, computational intelligence, and real-time data processing.

The high citation counts (reaching up to 3000) for these works indicate their impact in shaping AI-driven trading

strategies, with applications ranging from algorithmic trading to risk management. This survey underscores the

field's evolution from theoretical frameworks to practical implementations, driven by advancements in AI and

the increasing complexity of financial markets.

Figure 7. Most Relevant Words.

Page 212

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

The most relevant words (Figure 7) identified in the bibliometric analysis reveal key themes and focal points in

the application of Artificial Intelligence (AI) in stock market trading. The prominence of terms like stock,

market, and price underscores the central focus on financial markets and asset valuation. High-frequency words

such as model, learning, neural, machine, and deep highlight the dominance of machine learning and neural

networks in developing predictive and analytical tools for trading.

The recurrence of prediction, forecasting, and accuracy emphasizes the field's strong orientation toward

predictive analytics, where AI models are leveraged to anticipate market movements and optimize trading

strategies. Words like data, analysis, and time reflect the critical role of big data and real-time processing in

modern trading systems. Additionally, terms such as performance, results, and proposed indicate a research

emphasis on empirical validation and the development of novel methodologies.

This analysis suggests a robust interdisciplinary approach, combining financial theory with advanced AI

techniques, particularly deep learning and neural networks, to enhance decision-making and operational

efficiency in stock market trading. The high frequency of these relevant words aligns with the broader trends

observed in the most globally cited documents, reinforcing the field's focus on innovation, accuracy, and

practical applications.

Figure 8. Trend Topics.

The bibliometric analysis of trend topics (Figure 8) reveals a field that has evolved significantly from its early

theoretical foundations to today's sophisticated, data-driven approaches. The data (Figure 8) shows how neural

networks and deep learning architectures emerged in the late 1990s through terms like "mlcnns" and

"multilayered," laying the groundwork for modern algorithmic trading systems. These technical developments

coincided with the integration of classical financial models like Black-Scholes, demonstrating how AI gradually

merged with traditional quantitative finance. The persistent appearance of terms like "unexplainable" across

three decades highlights a fundamental challenge that remains unresolved - the interpretability of complex AI

models in an industry requiring transparency for regulatory compliance and risk management.

Current research trends emphasize hybrid approaches, as seen in terms like "neuro-fuzzy" and "evolutionary

algorithms," showing how researchers combine different AI techniques to handle market volatility and

uncertainty. The prominence of "back-propagation" and "self-organizing maps" underscores the continued

importance of both supervised and unsupervised learning methods. Meanwhile, emerging concepts like "smart-

" and references to specific regions ("Taiwan") suggest the field is expanding into decentralized finance and

localized applications. The analysis also reveals interesting interdisciplinary connections, with some terms

possibly indicating methodological borrowing from unrelated fields like civil engineering, though these may

represent noise requiring further verification.

Looking ahead, the field appears poised to address its enduring challenges around explainability while embracing

new opportunities in real-time processing, blockchain integration, and adaptive systems. The absence of very

recent innovations like transformer models in the dataset suggests either a focus on foundational works or the

need to expand the bibliometric corpus. What emerges most clearly is the field's trajectory from niche academic

research to a mature discipline that still must balance cutting-edge AI capabilities with the practical demands of

Page 213

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

financial markets - including regulatory scrutiny, risk management, and the need for models that can adapt to

ever-changing market conditions while remaining interpretable to human operators.

Figure 9. Clustering by Coupling. Coupling Measured by Abstracts and Cluster Labeling by Abstract Terms.

The coupling map cluster analysis (Figure 9) provides a nuanced perspective on the intellectual structure and

knowledge dynamics within AI-driven stock market trading research, revealing two primary yet interconnected

thematic clusters that shape the field's development. The first cluster (Group 1), characterized by the term

associations "stock-financial-data" with confidence levels spanning 36.8% to 53.2%, represents the fundamental

research pillar focused on the intersection of financial theory, data infrastructure, and basic predictive modeling.

This cluster's composition suggests it encompasses critical groundwork studies including financial data curation

and quality assessment (addressing challenges of noisy, high-frequency market data), feature extraction

methodologies (such as technical indicator engineering and fundamental factor analysis), and foundational

machine learning applications (like early neural networks and regression models for price prediction). With a

substantial impact score of 4.263 but relatively lower frequency (103 occurrences), this cluster appears to

represent high-value theoretical contributions that have enabled subsequent applied research, evidenced by its

strong centrality (0.39) indicating numerous conceptual linkages to other research domains. The cluster's

characteristic red hue (#E41A1C80) visually signifies its role as the foundational bedrock of the field.

The second cluster (Group 2), distinguished by the "stock-market-data" triad with stronger confidence levels

(52%-63.2%) and higher frequency (134 occurrences), embodies the field's evolution toward sophisticated

market applications and real-world implementation challenges. This cluster likely aggregates research on

algorithmic trading system architecture, market microstructure-informed AI models (including limit order book

dynamics analysis), high-frequency trading strategies, and regulatory-compliant execution frameworks. The

elevated confidence levels in term associations reflect more mature and well-established research relationships,

suggesting this cluster represents the field's current cutting edge where theoretical foundations are being stress-

tested against market realities. While showing slightly lower impact (4.16) than Group 1, its greater frequency

and strong centrality (0.411) position it as the dominant contemporary research paradigm. The cluster's blue

coloring (#377EB880) visually contrasts with Group 1 while maintaining chromatic harmony, symbolizing their

complementary rather than competing nature.

The relationship between these clusters reveals important insights about the field's knowledge progression.

Group 1's focus on "financial" theory versus Group 2's emphasis on "market" applications demonstrates a

maturation from abstract modeling to concrete implementation. The shared elements of "stock" and "data" in

both clusters form conceptual bridges, showing how fundamental financial data science (Group 1) continuously

feeds into advanced market applications (Group 2). The confidence level disparities suggest that while data-

financial connections remain somewhat variable (36.8%-53.2%), the data-market relationship has solidified into

a more consistent research paradigm (52%-63.2%). This may indicate that the field has reached consensus on

Page 214

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

market data applications while maintaining more diverse approaches to general financial data science. The near-

equational impacts only underscore the necessity of both theoretical and applied research; let supportive work

create breakthroughs, whereas practical applications drive innovation in reality, which in turn attracts attention

toward research illustrated by a high frequency finding in Group 2. The resulting view of the interrelationship

among all variables shows a healthy field, developing well, with strong fundamentals supporting a basic

discourse, as well as very high levels of both dimensions interpolating with strong links to surrounding research

areas due to substantial centrality numbers.

Figure 10. Clustering by Coupling. Network.

The bibliometric coupling map analysis (Figure 10) reveals two distinct yet interconnected AI application stock

market trading research clusters that collectively build the intellectual landscape of the field. Cluster 1 seems to

be the leading knowledge base with 85 high-impact papers boasting highly cited works such as Zhang Y (2020)

with an impressive normalized citation score of 163 and Asadi S (2012) at 150, demonstrating the enduring

effect of the earliest applications of machine earning in finance. This cluster has a strong representation in

prestigious AI journals (for example. Expert Systems with Applications, Applied Intelligence) and focuses on

core predictive modeling methodologies, such as neural networks (Kim K-J, 2012), hybrid intelligent systems

(Galeshchuk S, 2017), and data preprocessing methodologies. Going by the timeline, there is a strong trend of

research continuity in this cluster from 2012 to 2025, with recent works like Ashtiani MN (2023) and Aldhyani

THH (2022) outlining contemporary high-frequency trading challenges and explainable AI.

Cluster 2, while smaller (58 documents), contains several high-impact studies that push methodological

boundaries, evidenced by Picasso A (2019) with an exceptional citation score of 212 and Li Y (2022) at 128.

The group has a high representation in computational economics and Complex Systems readership (Chaos,

Computer Economics) indicating their interest in nonlinear market dynamics and adaptive trading systems. This

cluster gives importance to other kinds of science, like market microstructure applications and price prediction

using chaos theory, as shown by the contribution of such works as Motiwalla L (2000) and Wang H (2019).

Recent additions (2023-2025) demonstrate a growing interest in quantum computing applications (Kumar T,

2025) and sustainable finance (Munshi M, 2022).

The coupling patterns reveal important structural relationships. Cluster 1 serves as the methodological

foundation where core AI techniques are developed, while Cluster 2 represents their advanced market

applications. Numerous internal documentations among the clusters principally come from publications in

Expert Systems with Applications and IEEE Access, which appear in both of the groups. The normalized cites

(means) show Cluster 2 holds higher average impact (28.7 vs. 19.4 in the mean), indicating that applied market

research is receiving higher scholarly attention, whereas Cluster 1 exhibits more coherent throughput. Emerging

themes in both clusters (2024-2025 entries) signal that research interests are growing together in transformer

architectures, regulatory technologies, and ESG-based transaction systems, implying a richer development of all

being informed about the presence of more advanced ESG-AI applications. When analyzing the construction of

society and analyzing it from the premise that research is fundamentally sound and lively, scientific progress in

AI is continuously enhancing, inventing new marketing technologies alike.

Page 215

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 11. Co-Occurrence Network.

The co-word network analysis (Figure 11) gives a gross view of the conceptual framework and thematic

conventional form of AI applications concerning stock market trading research. Unlike Cluster 1, which appears

as a dense, tightly knit, and self-enclosed cluster of the network, it contains 48 nodes labeled as 'leading terms,'

with near-matching closeness centrality scores of 0.02 for all nodes. All terms have equal weight in the overall

structure-that is, maximum direct influence on other terms. Therefore, ontological resourcing forming Cluster 2

(detected by closeness centrality) can help on the concept of the use of AI in stock market prediction.

Top-relevant words were "stock", "market", and "data" that acquired respectively the greatest scores in PageRank

(which is 0.04, 0.039, and 0.034); this directly confirms the magnitude and significance of these keywords as

they aspectually serve as the anchor and the first thing to learn. With regard to methodological terms ("model,"

learning," "neural," "network") sharing equal importance as applicative terms ("prediction," "forecasting,"

"trading"), tech and practice are routinely stressed. Thus, a critical point ruled by neat methodical studies pivots

around how AI and big data are constructed in association with finance from either an innovative perspective or

an application. Interestingly, expressions "deep" and "lstm" have come to be special terms indicative of two AI

techniques integrated within the structure. More by the exclusionary evidence, the term "sentiment" is woven

with a relatively smaller thread (PageRank 0.008) pointing at the least of its integration into mainstream AI

trading research than the others.

The network comprises several telling characteristics for research. One is the strong up-front emphasis on

predictive modeling ("prediction," "forecasting," "predicting"), with almost all the layers having either one or

three of the terms in them. Second, while the library employs a lot of variations concerning neural network

approaches, it is pretty evenly spread, ranking for a solid combination of terms. They include "neural,"

"network," "deep," "LSTM." Third, the network takes proportionate consideration on the theoretical ("model,"

"algorithm") side as well as the practical ("price," "trading," "risk") influence. The fourth group of terms forming

the core appears to transform time-series into the new striking research focal points ("time," "series"). The

coherence index permits wording of the layers that mix these ingredients quite uniformly and efficiently, thus

forming nicely integrated knowledge domains where technical AI advances come in immediate connection with

financial market activities.

This analysis portrays AI in stock trading as a mature, cohesive field with clearly established conceptual

relationships, where new contributions typically build upon and connect to existing frameworks rather than

creating entirely separate research streams. The absence of isolated clusters or bridging terms suggests the field

has reached consensus on its core paradigms while continuing to develop specialized techniques within this

unified framework.

Page 216

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 12. Thematic map - Abstracts.

The bibliometric thematic mapping (Figure 12) for AI application in stock trading suggests a mature yet

developing field divided into three key research domains, each quite distinct in the knowledge landscape. The

node "market" is observed to be the very active core of the field, centered around high centralities (0.943) and

density (0.805) with 144,582 occurrences, translating this research area as the most advanced and in-field as

well. This driving theme has been responsible for a lot of starting work in fields such as algorithmic trading

systems, real-time market analytics, and AI-based price prediction models, where refined machine learning

skills, like those used to address complex market issues. The other cluster, "stock," has a centrality of 0.866 and

a density of 0.726, with an emphasis more much on other applications under the rubric of equity, with portfolio

optimization and risk assessment, to show how basic AI principles could be adjusted to solve concrete financial

issues.

The "artificial" cluster (centrality 0.405, density 0.512) represents the field's methodological backbone,

containing fundamental AI/ML algorithms and their financial adaptations. While less prominent in terms of

research volume (35,565 occurrences) and centrality, its moderate density indicates established technical rigor,

serving as the essential substrate supporting applied innovations. The hierarchical ranking of these clusters

reveals a clear knowledge diffusion pathway. core AI principles evolve in the "artificial" domain, are refined for

financial instruments in the "stock" cluster, and ultimately deploy at scale in comprehensive "market"

applications.

Notably absent are specialized niche themes (high-density, low-centrality), suggesting the field currently

prioritizes practical applicability over isolated theoretical advancements. This is further evidenced by the

concentration of themes along the centrality axis, where even basic themes like "stock model models" maintain

high relevance despite lower developmental activity. The emerging/declining status of "artificial system" themes

hints at both opportunities and challenges in developing integrated AI trading architectures, potentially

representing the next frontier for research investment.

The field exhibits healthy knowledge transfer mechanisms between theory and practice, though the moderate

density of methodological research suggests opportunities for deeper technical innovation to support next-

generation applications. Future development would benefit from cultivating specialized sub-domains (e.g.,

explainable AI for trading, quantum financial models) while maintaining the strong applied focus that currently

drives progress. This dual emphasis on both advancing core methodologies and solving real-world market

problems positions AI in stock trading as a exemplar of translational financial research, where theoretical insights

rapidly transform into market-ready solutions.

Page 217

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 13. Thematic Map. Network.

The thematic map terms analysis (Figure 13) reveals a well-structured research landscape in AI-driven stock

market trading, organized into three distinct but interconnected clusters that reflect the field's theoretical

foundations, methodological approaches, and practical applications. The "stock" cluster (6,583 occurrences)

emerges as the most specialized domain, focusing intensely on predictive modeling techniques with terms like

"prediction," "neural network," "LSTM," and "forecasting" demonstrating the field's strong emphasis on time-

series analysis and deep learning applications. This cluster shows particular sophistication in its technical

vocabulary, featuring specialized terms like "ARIMA" (385 occurrences), "CNN" (446), and "SVM" (396) that

reveal ongoing innovation in combining traditional statistical methods with modern AI architectures. The

presence of evaluation metrics like "RMSE" (516) and performance terms like "accuracy" (2,847) underscores

the quantitative rigor characterizing this research strand.

Cluster 2, labeled "artificial" (1,998 occurrences), serves as the conceptual bridge between AI theory and

financial practice, containing foundational terms like "intelligence," "optimization," and "decision-making"

alongside application-oriented concepts such as "portfolio management" and "trading strategies." This cluster's

unique value lies in its inclusion of emerging paradigms, evidenced by terms like "reinforcement learning" (552)

and "text mining" (387), which point to cutting-edge applications of AI in market sentiment analysis and adaptive

trading systems. The cluster also captures the field's methodological diversity through terms like "detection"

(2,072 betweenness centrality) and "graph theory" (10,647), suggesting growing interest in anomaly detection

and network-based market analysis approaches.

The "market" cluster dominates in both size and centrality, with "market" itself appearing 6,647 times alongside

high-frequency companion terms like "data" (5,554), "learning" (5,228), and "financial" (4,640). This cluster

embodies the field's applied dimension, connecting AI methodologies to real-world market contexts through

terms like "trading strategies" (1,004), "investor behavior" (1958), and "market trends" (1105). Notably, it

incorporates both traditional approaches ("statistical methods" - 849) and contemporary innovations ("deep

learning" - 2,443), reflecting the field's evolutionary trajectory. The cluster's substantial representation of

evaluation terms ("effective" - 935, "empirical" - 898) highlights the strong empirical orientation of market-

focused AI research.

Several key insights emerge from the betweenness centrality metrics, which reveal critical bridging terms

connecting these clusters. "IEEE" (15.083) and "ANN" (1,432.5) serve as important conceptual links between

technical and applied research, while "futures" (18,544) unexpectedly emerges as a crucial connector, likely due

to its dual role as both a financial instrument and a temporal forecasting concept. The analysis also identifies

underdeveloped areas, with "sentiment analysis" (893) and "fuzzy systems" (421) showing relatively low

penetration despite their potential relevance, suggesting opportunities for future research expansion. The overall

thematic structure portrays a field that has achieved strong integration between AI methodologies and financial

applications while maintaining robust theoretical foundations, with clear pathways for knowledge transfer from

fundamental research ("artificial") to specialized prediction models ("stock") and ultimately to market

implementations ("market").

Page 218

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 14. Thematic Evolution.

The thematic evolution analysis (Figure 14) reveals a remarkable transformation in AI-driven stock market

trading research across five decades, demonstrating both the field's dynamic progression and its enduring

foundational themes. The early period (1971-2000) established fundamental building blocks, with "data,"

"neural," and "stock" emerging as core themes that would branch significantly in subsequent decades. The 2001-

2008 period marked a pivotal transition where neural network applications exploded in sophistication (weighted

inclusion index 0.4), branching into forecasting, decision systems, and complex market analysis, while stock-

related research evolved from basic modeling to incorporate advanced concepts like nonlinear analysis (0.51)

and technical indicators. This era saw the crystallization of key methodologies, with neural networks

transitioning from theoretical tools to practical applications in market prediction (0.63 inclusion from neural to

stock).

The 2009-2016 period witnessed substantial thematic diversification, as evidenced by the 0.54 inclusion index

from neural to financial themes, incorporating machine learning, fuzzy systems, and optimization techniques.

Stock research during this phase achieved remarkable maturity (0.7 inclusion) by integrating temporal analysis,

hybrid models, and sophisticated evaluation metrics. The most recent phases (2017-2024 and 2025) demonstrate

the field's accelerated convergence with cutting-edge AI, where financial and market themes incorporated deep

learning (0.77) and sentiment analysis, while stock-specific research embraced LSTM (appearing in 1,674

occurrences) and convolutional networks. The emergence of "based" as a dominant 2025 theme (2,944

occurrences) with 0.58 inclusion from market research reflects the field's maturation into application-focused

studies built on established AI foundations. Current trajectories show increasing emphasis on real-world

implementation challenges (sentiment analysis, reinforcement learning) alongside technical refinement

(attention mechanisms, transformer architectures), while maintaining strong connections to core financial

concepts like volatility and portfolio optimization. This evolution portrays a field that has successfully

transitioned from theoretical exploration to robust application while continuously absorbing advancements from

broader AI research, with current work poised to tackle increasingly complex market dynamics through

sophisticated, explainable AI systems.

In summary, the research paper identifies several critical gaps in the application of AI to stock market trading

through its comprehensive bibliometric analysis. A fundamental challenge highlighted is the issue of

"unexplainable AI," where advanced models like deep learning and LSTM networks demonstrate strong

predictive performance but lack transparency in their decision-making processes, creating barriers to adoption

in regulated financial markets. This interpretability gap is compounded by models' inability to adapt effectively

to sudden market volatility, as most systems remain overly dependent on historical data patterns rather than

incorporating real-time adjustment mechanisms. The analysis also reveals significant underutilization of

alternative data sources, with bibliometric indicators showing low integration of sentiment analysis and other

unstructured data streams compared to traditional quantitative methods. While hybrid approaches like neuro-

fuzzy systems and evolutionary algorithms show promise, the research indicates these remain underexplored

relative to conventional neural network architectures. The paper further identifies ethical and regulatory

challenges, particularly the absence of robust frameworks for addressing algorithmic biases and aligning AI

systems with ESG principles. Technical limitations in real-time processing capabilities, especially for high-

frequency trading applications, and the embryonic state of quantum computing implementations in financial

contexts are noted as additional constraints.

Page 219

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

To address these gaps, the paper proposes several key future research directions. Enhancing model

interpretability through Explainable AI (XAI) techniques like SHAP and LIME is emphasized as crucial for

building trust and regulatory compliance. Developing real-time adaptive models using reinforcement learning

and online learning approaches could improve responsiveness to market fluctuations. The integration of

multimodal data sources, including sentiment analysis and satellite imagery, is recommended to complement

traditional market data. The research highlights hybrid architectures as a promising avenue for balancing

accuracy with interpretability, while also calling for greater focus on ethical AI development aligned with ESG

objectives. Interdisciplinary collaboration among finance, computer science, and behavioural economics is

identified as essential for advancing the field, along with investments in regulatory technology (RegTech) to

ensure that AI trading systems meet compliance requirements. These recommendations are grounded in the

paper's bibliometric findings, including thematic analyses showing the persistence of "unexplainable AI"

challenges and the emerging but underdeveloped status of hybrid and real-time modeling approaches. The

conclusions drawn directly reflect the authors' identification of current limitations and their proposed roadmap

for future research in AI-driven financial decision-making.

The research paper acknowledges quantum AI as an emerging but still nascent area in stock market trading

applications. Quantum computing appears in Cluster 2 of the coupling analysis (Figure 10), linked to "adaptive

trading systems" and "sustainable finance," but with low frequency. The "Thematic Evolution" (Figure 14)

shows quantum AI as an emerging 2025 theme (term. "quantum," occurrence. 2,944) but with low inclusion

indices (0.58), indicating it’s not yet mainstream. While the research includes quantum computing in its

bibliometric analysis, the coverage remains relatively surface-level due to the field's early developmental stage.

The paper identifies quantum AI's potential applications in three key areas. portfolio optimization (where

quantum algorithms could solve complex problems faster than classical methods), high-frequency trading

(through quantum machine learning's ability to process vast datasets exponentially quicker), and risk modeling

(using quantum-enhanced Monte Carlo simulations). However, the analysis reveals significant gaps in current

quantum AI research for financial applications, including hardware scalability limitations with current NISQ

devices, a lack of empirical studies on hybrid classical-quantum models, and absent regulatory frameworks for

quantum-powered trading systems. The paper positions quantum AI as part of future research directions rather

than current mainstream applications, suggesting areas like quantum neural networks and quantum data encoding

as promising avenues. This limited treatment reflects the field's immaturity - evidenced by quantum terms

appearing with low frequency in the bibliometric analysis and not ranking among the most relevant

keywords.Use either SI (MKS) or CGS as primary units. (SI units are strongly encouraged.) English units may

be used as secondary units (in parentheses). This applies to papers in data storage. For example, write “15

Gb/cm2 (100 Gb/in2).” An exception is when English units are used as identifiers in trade, such as “3½-in disk

drive.” Avoid combining SI and CGS units, such as current in amperes and magnetic field in oersteds. This often

leads to confusion because equations do not balance dimensionally. If you must use mixed units, clearly state

the units for each quantity in an equation.

Theoretical Framework

Figure 15 shows the core financial theories on which this study is based to provide a backdrop against the use of

AI in stock market trading. The EMH provides the fundamental assumption that markets efficiently incorporate

all available information into prices. AI systems challenge the EMH by detecting hidden patterns of temporary

market inefficiencies through advanced machine learning methodologies, especially deep learning systems such

as LSTMs and CNNs. This has been seen with bibliometric maps where terms like "neural networks" and

"forecasting" take center stage in the studied literature. Behavioral Finance Theory adds into the mix

psychological factors and cognitive biases that affect investors' behavior. AI has started implementing this via

sentiment analysis of news and social media; however, the relatively low occurrence of the word "sentiment" in

co-occurrence networks would imply that this area is still rather unexplored. Bridging these views, the Adaptive

Market Hypothesis envisages markets as evolving systems in which the participants continuously adapt. This

stands well for how AI works, driven by machine learning, especially reinforcement learning systems that allow

agents to change with market conditions. From 2017 to 2025, these developments are very clear in the spotlight

of "real-time" and "adaptive" modeling. Altogether, these theories justify the application of AI in trading while

also spotlighting critical research gaps, especially concerning behavioral integration and model interpretability.

Page 220

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

Figure 15. Theoretical Framework.

Conceptual Framework

The conceptual framework considers AI-enabled stock trading to consist of four main components (Figure 16),

all interlinked. Transformation data consists of both structured data of historical prices and fundamentals

dominating current research (depicted in Cluster 1 focusing on "stock-financial-data") and neglected

unstructured data such as news and satellite images, for example. Processing is performed by two methodological

approaches: one being predictive modeling with dominant architectures like LSTMs and CNNs, and an adaptive

learning system that adjusts itself as markets change and, as such, is exploited with models that blend statistical

and AI-based ones. The mainstay of validation relies on traditional measures like RMSE or accuracy, but real

stress-testing of models in times of high volatility has yet to be developed. The output layer is where algorithmic

trading strategies (the dominant cluster in bibliometrics is "market") and risk management tools are indeed

produced, yet governance criteria on the other hand-has increasingly been incorporated, such as ESG alignment

and explainability requirements (XAI). Processes in feedback foster continual system improvement through real-

time retraining of models and adding alternative data sources, thereby evolving research to define and eliminate

current gaps like incorporating quantum AI and stronger linkages to behavioral finance. Hence, this framework

maps the current AI trading ecosystem and identifies avenues for future development, arranging the very strong

technical turf of the field against the budding questions of ethics and interdisciplinary thoughts.

CONCLUSION

This comprehensive bibliometric study presents a unique and systematic evaluation of artificial intelligence

applications in stock market trading by analyzing an unprecedented set of 9,088 documents over 54 years (1971-

2025). What makes this uniquely novel is its multifaceted approach which combines bibliometric quantitative

analysis with qualitative theoretical synthesis, showing both the evolution and future directions of the field. The

study recognizes three main research clusters - market applications, predictive modeling, and AI methodologies

- beyond which it portrays how neural networks and LSTM models have commandeered developments in recent

years, with transformer architectures and explainable AI as emerging frontiers.

This research goes somewhat further than just tracking developments and contributes novel insights. Firstly, it

exposes sore yet understudied paradoxes in the field, especially the tension between model sophistication and

interpretability (the "explainability-adaptability paradox"). Secondly, it exposes the major geographical

disparities in focus, with East Asian institutions taking the forefront in algorithmic development, while Western

research is at the rear in behavioral finance integration. Thirdly, the study pioneers an original four-layer

conceptual framework that correlates technical AI capabilities with the actual needs of trading, regulatory

requirements, and ethical considerations.

What enables this research to stand out is that it manages to accomplish a connection between theory and

practice. While the study grounds its findings in financial theories such as Efficient Market Hypothesis and

Adaptive Markets Theory, it also, in turn, addresses practical-level problems like volatility response and

regulation compliance so that it ends up with possible diagnostic and prescriptive measures. Especially from the

rigorous trend analysis, identification of quantum AI and ESG-aligned systems as next frontiers will be of much

value to the scientific community concerning the direction of their research.

This work yields not only a different methodological approach capable of integrating perspectives not previously

Page 221

www.rsisinternational.org

INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)

ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025

interlinked and fills in gaps in knowledge by compiling the most complete intellectual history of AI in trading

today but also responsible avenues for going forward. The unique blend of historical comprehensiveness,

theoretical innovation, and practical applicability marks this work as a must-have for academics developing new

AI techniques, practitioners implementing trading systems, and policymakers shaping financial regulations in

the AI age.

REFERENCES

1. Alonso-Garcia, J., Pablo-Martí, F., & Nunez-Barriopedro, E. (2021). Omnichannel Management in a

B2B Context: Concept, Research Agenda and Bibliometric Review. International Journal of Industrial

Engineering and Management, 12(1), 37-48. https://doi.org/10.24867/IJIEM-2021-1-275

2. Aria, M., & Cuccurullo, C. (2017). bibliometrix: An R-tool for comprehensive science mapping analysis.

Journal of Informetrics, 11(4), 959-975. https://doi.org/10.1016/j.joi.2017.08.007

3. Bar-Ilan, J. (2008). Informetrics at the beginning of the 21st century—A review. Journal of Informetrics,

2(1), 1-52. https://doi.org/10.1016/j.joi.2007.11.001

4. Bar-Ilan, J. (2010). Citations to the 'Introduction to informetrics' indexed by WOS, Scopus and Google

Scholar. Scientometrics, 82(3), 495-506. https://doi.org/10.1007/s11192-010-0185-9

5. Bastian, M., Heymann, S., & Jacomy, M. (2009). Gephi: an open source software for exploring and

manipulating networks. Proceedings of the International AAAI Conference on Web and Social Media,

3(1), 361-362.

6. Donthu, N., Kumar, S., Mukherjee, D., Pandey, N., & Lim, W. M. (2021). How to conduct a bibliometric

analysis: An overview and guidelines. Journal of Business Research, 133, 285-296.

https://doi.org/10.1016/j.jbusres.2021.04.070

7. Egghe, L., & Rousseau, R. (1990). Introduction to informetrics. Quantitative methods in library,

documentation and information science. Elsevier Science Publishers.

8. Fahimnia, B., Sarkis, J., & Davarzani, H. (2015). Green supply chain management: A review and

bibliometric analysis. International Journal of Production Economics, 162, 101-114.

https://doi.org/10.1016/j.ijpe.2015.01.003

9. Fan, D., Chen, L., & Zhang, F. (2022). Advancing literature review methodology through rigour,

generativity, scope and transparency. International Journal of Management Reviews, 24(2), 171-180.

https://doi.org/10.1111/ijmr.12268

10. Hicks, D., Wouters, P., Waltman, L., De Rijcke, S., & Rafols, I. (2015). Bibliometrics: the Leiden

Manifesto for research metrics. Nature, 520(7548), 429-431. https://doi.org/10.1038/520429a

11. Huffman, M. D., Baldridge, A. S., Bloomfield, G. S., Colantonio, L. D., Prabhakaran, D., Ajay, V. S., ...

& Vedanthan, R. (2013). Global cardiovascular research output, citations, and collaborations: a time-

trend,

https://doi.org/10.1371/journal.pone.0083440

12. Kunisch, S., Menz, M., Bartunek, J. M., Cardinal, L. B., Denyer, D., & Smith, W. K. (2023). Review

research as scientific inquiry. Organizational Research Methods, 26(1), 3-45.

https://doi.org/10.1177/1094428121999094

bibliometric

analysis

(1999-2008).

PloS

One,

8(12),

e83440.

13. Liu, X., Zhan, F. B., Hong, S., & Niu, B. (2012). A bibliometric study of earthquake research: 1900-

2010. Scientometrics, 92(3), 747-765. https://doi.org/10.1007/s11192-011-0599-z

14. Paul, J., Lim, W. M., O'Cass, A., Hao, A. W., & Bresciani, S. (2021). Scientific procedures and rationales

for systematic literature reviews (SPAR‐4‐SLR). International Journal of Consumer Studies, 45(4), O1-

O16. https://doi.org/10.1111/ijcs.12695

15. Pellegrino Jr, D. A. (2011). Interactive visualization systems and data integration methods for supporting

discovery in collections of scientific information. (Doctoral dissertation, Drexel University).

16. Wolfram, D. (2003). Applied informetrics for information retrieval research. Greenwood Publishing

Group.

17. Zaugg, H., West, R. E., Tateishi, I., & Randall, D. L. (2011). Mendeley: Creating communities of

scholarly inquiry through research collaboration. In Proceedings of the 2011 iConference (pp. 32-39).

https://doi.org/10.1145/1940761.1940767

18. Zhuang, Y., Liu, X., & Porter, A. L. (2013). Global remote sensing research trends during 1991-2010: a

bibliometric analysis.Scientometrics,6(1),203-219. https://doi.org/10.1007/s11192-012-0918

Page 222

www.rsisinternational.org