AI Holds Strong Potential to Transform Stock Market Forecasting in Nepal

Artificial Intelligence (AI) is emerging as a promising technology for stock market forecasting, with growing potential to support investment decisions and transform the way investors analyze the Nepal Stock Exchange (NEPSE). A study published in the Journal of Business and Social Sciences Research (JBSSR) has examined the use of AI in stock market prediction, highlighting the potential of machine learning and deep learning models to address the challenges created by the highly volatile, complex, and nonlinear nature of stock markets. Predicting stock prices has always been difficult because markets constantly respond to new information, economic conditions, investor behavior, and unexpected events. Stock prices are non-stationary and volatile, making it difficult for traditional statistical methods to accurately forecast future movements. The study notes that conventional financial theories, including the Random Walk Model and Efficient Market Hypothesis, suggest that stock prices are difficult to predict consistently because new information is rapidly reflected in market prices. However, research in behavioral finance and social science has challenged the idea that markets are completely unpredictable. This has created opportunities for advanced computational techniques to identify patterns and relationships that may not be easily detected through conventional analysis. AI has therefore attracted increasing attention in financial markets because it can process large amounts of data, recognize complex patterns, and generate forecasts without relying entirely on predetermined relationships. Among the most widely studied AI technologies for stock market prediction are Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network (CNN) models. These deep learning techniques are particularly useful for analyzing financial time-series data and identifying relationships between past and present market movements. Research focusing on NEPSE by Pokheral et al. (2022) compared LSTM, GRU, and CNN models using 16 different predictors and multiple datasets. The findings showed that LSTM achieved better prediction accuracy than both GRU and CNN. The result is significant for Nepal because LSTM is designed to retain important information over longer periods, making it suitable for analyzing sequential data such as stock prices and market indices. Another emerging technology in Nepal is the use of Graph Neural Networks (GNNs). Unlike conventional models that may analyze stocks and indicators separately, graph-based AI can represent financial information as interconnected networks. This allows the model to examine relationships between different market variables and identify complex dependencies. A 2025 study by Bajracharya et al. applied GNNs to NEPSE data and compared Graph Convolutional Networks (GCN) with Graph Attention Networks (GAT). The research found that GCN performed better than GAT in learning complex market relationships and long-term dependencies. This suggests that graph-based models could offer another promising direction for improving stock market forecasting in Nepal. Globally, AI has moved beyond academic research and is increasingly being used in practical financial applications. Financial institutions are using AI and machine learning for algorithmic trading, portfolio management, risk assessment, fraud detection and high-frequency trading. AI can analyze large volumes of information at high speed and support investors in identifying market trends and managing risks. Recent research has also examined the integration of large language models (LLMs) with machine learning in investment strategies. This indicates that AI technologies such as LLMs could provide additional information and analytical capabilities, although their effectiveness depends on the type of forecasting strategy and data being used. The global experience demonstrates that AI can become more than a forecasting tool. It can potentially support the entire investment decision-making process. Despite the promising research results, AI adoption in Nepal's stock market remains at an early stage. The major gap is between academic research and real-world application. Most studies conducted in Nepal have focused on comparing the performance of different AI models rather than applying them directly to live trading and investment decisions. There is therefore a need for further research into how AI models perform under actual NEPSE market conditions. Real-time testing, larger datasets, and practical investment applications could provide a clearer understanding of AI's effectiveness. The development of AI-based investment tools may also require improvements in data quality, technological infrastructure, investor awareness and regulatory frameworks. While AI can improve forecasting and analysis, it cannot eliminate the risks associated with stock market investment. Market movements can be influenced by unexpected political developments, economic shocks, policy changes, company announcements and investor sentiment. Such events may not always be accurately predicted from historical data. For this reason, the study emphasizes the importance of combining AI-based analysis with human expertise and judgment. AI can provide data-driven insights, while investors can use financial knowledge and broader market understanding to make final decisions. The growing body of research suggests that AI has considerable potential to become an important part of Nepal's capital market. Models such as LSTM and GCN have already demonstrated promising results in analyzing and forecasting NEPSE movements. The next challenge is to move beyond academic experiments and explore practical applications for investors, brokers, portfolio managers, and financial institutions. If supported by reliable data, appropriate technology and effective regulation, AI could help Nepal's stock market become more data-driven and efficient. For NEPSE, the future opportunity lies not simply in replacing traditional investment analysis with AI, but in combining advanced technology with human expertise to make faster, better-informed, and more disciplined investment decisions. Source: https://www.nepjol.info/index.php/jbssr/article/view/98984