Search results

This study aims to critically evaluate the competitiveness of Transformer-based models in financial forecasting, specifically in the context of stock realized volatility forecasting. It seeks to challenge and extend upon the assertions of Zeng et al. (2023) regarding the purported limitations of these models in handling temporal information in financial time series.

Employing a robust methodological framework, the study systematically compares a range of Transformer models, including first-generation and advanced iterations like Informer, Autoformer, and PatchTST, against benchmark models (HAR, NBEATSx, NHITS, and TimesNet). The evaluation encompasses 80 different stocks, four error metrics, four statistical tests, and three robustness tests designed to reflect diverse market conditions and data availability scenarios.

The research uncovers that while first-generation Transformer models, like TFT, underperform in financial forecasting, second-generation models like Informer, Autoformer, and PatchTST demonstrate remarkable efficacy, especially in scenarios characterized by limited historical data and market volatility. The study also highlights the nuanced performance of these models across different forecasting horizons and error metrics, showcasing their potential as robust tools in financial forecasting, which contradicts the findings of Zeng et al. (2023)

This paper contributes to the financial forecasting literature by providing a comprehensive analysis of the applicability of Transformer-based models in this domain. It offers new insights into the capabilities of these models, especially their adaptability to different market conditions and forecasting requirements, challenging the existing skepticism created by Zeng et al. (2023) about their utility in financial forecasting.

This paper aims to analyze the role and performance of different artificial intelligence (AI) algorithms in forecasting future movements in the main indices of the world’s largest stock exchanges.

Drawing on finance-based theory, an empirical and experimental study was carried out using four AI-based models. The investigation comprised training, testing and analysis of model performance using accuracy metrics and F1-Score on data from 34 indices, using 9 technical indicators, descriptive statistics, Shapiro–Wilk, Student’s t and Mann–Whitney and Spearman correlation coefficient tests.

All AI-based models performed better than the markets' return expectations, thereby supporting financial, strategic and organizational decisions. The number of days used to calculate the technical indicators enabled the development of models with better performance. Those based on the random forest algorithm present better results than other AI algorithms, regardless of the performance metric adopted.

The study expands knowledge on the topic and provides robust evidence on the role of AI in financial analysis and decision-making, as well as in predicting the movements of the largest stock exchanges in the world. This brings theoretical, strategic and managerial contributions, enabling the discussion of efficient market hypothesis (EMH) in a complex economic reality – in which the use of automation and application of AI has been expanded, opening new avenues of future investigation and the extensive use of technical analysis as support for decisions and machine learning.

The AI algorithms' flexibility to determine their parameters and the window for measuring and estimating technical indicators provide contextually adjusted models that can entail the best possible performance. This expands the informational and decision-making capacity of investors, managers, controllers, market analysts and other economic agents while emphasizing the role of AI algorithms in improving resource allocation in the financial and capital markets.

The originality and value of the research come from the methodology and systematic testing of the EMH through the main indices of the world’s largest stock exchanges – something still unprecedented despite being widely expected by scholars and the market.

The accuracy of budgeting is important to fulfilling its various roles. The aim of this study is to examine perceptions of budgeting accuracy in UK universities and to identify and understand the factors that influence them.

A mixed methods research design comprising a questionnaire survey (84 responses, = 51.5%) and 42 semi-structured, qualitative interviews is employed.

The findings reveal that universities tend to be conservative in their budgeting, although previous financial difficulties, the attitude of the governing body and the need to convince lenders that finances are being managed competently might lead to a greater emphasis on a “realistic” rather than cautious budget. Stepwise multiple regression identified four significantly negative influences on perceived budgeting accuracy: the difficulty of forecasting student numbers; difficulties associated with allowing unspent balances to be carried forward; taking a relatively long time to prepare the budget; and the institution’s level of financial surplus. The interviews are drawn upon to both explain and elaborate on the statistical findings. Forecasting student numbers and associated fee income emerges as a particularly challenging and complex issue.

Our regression analysis is cross-sectional and therefore based on correlations. Furthermore, the research could be developed by investigating the views of other parties as well as repeating the study in both the UK and overseas.

Implications for university management follow from the four factors identified as significant influences upon budget accuracy. These include involving the finance department in estimating student numbers, removing or controlling the carry forward of unspent funds, and reducing the length of the budget cycle.

The first study to examine the factors that influence the perceived accuracy of universities’ budgeting, this paper also advances understanding of budgeting accuracy more generally.

In today's advanced economy, there is a broader presence of information revolution, such as artificial intelligence (AI). AI primarily drives modern banking, leading to innovative banking channels, services and solutions disruptions. Thus, this chapter intends to determine AI's place in contemporary banking and stock market trading.

Stock market forecasting is hampered by the inherently noisy environments and significant volatility surrounding market trends. There needs to be more research on the mantle of AI in revolutionising banking and stock market trading. Attempting to bridge this gap, the present research study looks at the function of AI in banking and stock market trading.

The researchers have synthesised the literature pool. They undertook a systematic review and meta-synthesis method by identifying the major themes and a systematic literature review aided in the critical analysis, synthesis and mapping of the body of existing material.

The study's conclusions demonstrated the efficacy of AI, which has played a robust role in banking and finance by reducing risk and operational costs, enabling better customer experience, improving regulatory complaints and fraud detection and improving credit and loan decisions. AI has revolutionised stock market trading by forecasting future prices or trends in financial assets, optimising financial portfolios and analysing news or social media comments on the assets or firms.

AI's debut in banking and finance has brought sea changes in banking and stock market trading. AI in the banking industry and capital market can provide timely and apt information to its customers and customise the products as per their requirements.

Motivated by empirical features that characterize cryptocurrency volatility data, the authors develop a forecasting strategy that can account for both model uncertainty and heteroskedasticity of unknown form. The theoretical investigation establishes the asymptotic optimality of the proposed heteroskedastic model averaging heterogeneous autoregressive (H-MAHAR) estimator under mild conditions. The authors additionally examine the convergence rate of the estimated weights of the proposed H-MAHAR estimator. This analysis sheds new light on the asymptotic properties of the least squares model averaging estimator under alternative complicated data generating processes (DGPs). To examine the performance of the H-MAHAR estimator, the authors conduct an out-of-sample forecasting application involving 22 different cryptocurrency assets. The results emphasize the importance of accounting for both model uncertainty and heteroskedasticity in practice.

This paper aims to conduct a comparative analysis of the impact of market uncertainty on the degree of accuracy and bias of analysts' earnings forecasts versus four model-based earnings forecasts.

The study employs panel regression analysis on a sample of Egyptian listed companies from 2005 to 2022 to examine the impact of market uncertainty on the accuracy and bias of each type of earnings forecast.

The empirical analysis reveals that market uncertainty significantly affects analysts’ earnings forecast accuracy and bias, while model-based earnings forecasts are less affected. Furthermore, the Earnings Persistence and Residual Income model-based earnings were found to be superior in terms of exhibiting the least susceptibility to the impact of market uncertainty on their forecast accuracy and biasness levels, respectively.

The findings have important implications for stakeholders within the financial realm, including investors, financial analysts, corporate executives and portfolio managers. They emphasize the importance of considering market uncertainty when formulating earnings forecasts, while concurrently highlighting the potential benefits of using alternative forecasting methods.

To our knowledge, the influence of market uncertainty on analysts' earnings forecast accuracy and bias in the MENA region, particularly in the Egyptian market, remains unexplored in existing research. Additionally, this paper contributes to the existing literature by pinpointing the forecasting method, specifically distinguishing between analysts-based and model-based approaches, whose predictive quality is less adversely impacted by market uncertainty in an emerging market.

This study aims to predict stock market crises in the Middle East North Africa (MENA) regions by leveraging the nonlinear autoregressive neural network with exogenous inputs (NARX) model with two measures of investor sentiment: the ARMS indicator and Google Trends' search volume of positive and negative words.

Employing a novel approach, this study utilizes the NARX model with ten neurons in the hidden layer and the Levenberg–Marquardt training algorithm. It evaluates model performance through learning, validation and test errors, as well as correlation analysis between predicted and actual crises.

The NARX model, incorporating investor sentiment, has proven to be a reliable tool for forecasting crises, helping market participants understand data complexity and avoid crisis consequences. The divergence in how investors interpret market news, with some focusing solely on negative developments and others valuing positive outcomes, highlights the predictive nature of the optimistic and pessimistic sentiments captured by the model.

This study advocates for integrating behavioral approaches into stock market crisis prediction, highlighting the significance of investor sentiment and deep learning. It advances crisis mechanism understanding and opens avenues in behavioral finance. Integration of these findings into finance and economics education could enhance students' risk understanding and mitigation strategies.

The adoption of NARX models, incorporating investor sentiment, empowers market participants to proactively manage crises, adjust strategies, enhance asset protection and make informed decisions. These models enable them to minimize losses, maximize returns and diversify portfolios effectively in response to market fluctuations. These insights also guide policymakers such as governments, regulatory institutions and financial organizations in formulating crisis prevention and mitigation policies, bolstering economic and financial stability.

This research reduces economic uncertainty, safeguards individuals' savings and investments and promotes a stable financial climate.

This study is one of the first attempts to demonstrate the detection and prediction of stock market crises, specifically in the MENA stock market, using the NARX model. It offers a robust forecasting model using machine learning and investor sentiment, providing decision-making support for investment strategies and policy development aimed at enhancing financial and economic stability.