The increase in financial illicit activities between South Africa and Zimbabwe borders, which are estimated to lose USD 3.1 billion yearly (SARB, 2024; RBZ, 2023), motivates an AI application that assists the traditional techniques. This research implements FALCON (Financial Anomaly Detection via Contextual Learning Optimized Network), a hybrid architecture of transformer–GNN models developed by South Africa’s Financial Intelligence Centre (FIC) as well as Zimbabwe’s Reserve Bank (RBZ) and SWIFT. By employing temporal transaction pattern (TimeGAN) and entity mapping based on graphs (GraphSAGE), FALCON detected money laundering techniques with 98.7% accuracy, which surpasses Random Forest (72.1%) and human auditors (64.5%). Additionally, it also lowered the false positives to 1.2% (AUC-ROC: 0.992). After testing the model on 1.8 million transactions (falsified South Africa Central Bank (SARB) CTRs and RBZ STRs) and Ethereum blockchain (Etherscan.io), FALCON uncovered USD 450 million intentionally hidden funds that flowed through 23 shell companies. The model's XAI (SHAP) outputs explainable artificial intelligence are compliant with FATF, meaning no legislative exorbitant scrutiny, which is the requirement to create evidence that can stand in a court of law, which in trial phases had a 92% acceptance rate. The main innovations are the model's capabilities of extending beyond borders, which identifies the SARB-RBZ gap in transactions with 94% precision, masking sensitive (differential privacy, ε=1.2) data compliant with the General Data Protection Regulation (GDPR), and processing 2M transactions per second on AWS Graviton3, achieving real-time scalability. As the first AI framework designed for Southern Africa’s financial ecosystems, the FALCON AI Framework serves as the gold-standard claimable framework for ethical AI in emerging economies since it is entirely validated on public data. It can be used immediately for Central Bank Digital Currency supervision.

This study investigates the forecasting accuracy of stock prices and indices in two developed countries, the UK and the US, and two developing countries, China and India, covering a range of market capitalisations, spanning large-, mid-, and small-cap companies and indices. This study is based on data collected over a period of 13 years. I deploy Particle-Swarm-Optimised Radial Basis Function Neural Networks (i.e., PSO-RBFNNs) and compare their performance with that of the traditional RBF Neural Network (i.e., RBFNN) model and two benchmark econometric models: the ARIMA model and the Holt–Winters model. This study employs technical indicators. The results show that in developed countries, econometrics models often perform better than neural network models, except for the US small-cap stock index S&P 600, while the PSO-RBFNN model outperforms the traditional RBFNN model in the vast majority of cases due to the optimisation of the parameters of the RBFNN model by the particle swarm optimisation algorithm. In emerging market data, neural network models outperform econometric models, while PSO-RBF performs better than or similarly to RBF in the vast majority of cases. As observed from the results, neural networks are able to provide better predictive performance for data containing complex nonlinear patterns and relationships to some extent.

Despite the theoretical limitations posed by the Efficient Market Hypothesis, technical indicators remain widely used in high-frequency trading (HFT). However, their effectiveness at minute-level frequencies, where market microstructure effects dominate, remains underexplored. This study evaluates the role of traditional technical indicators in Random Forest Regression (RFR) models using minute-level SPY data across 13 distinct configurations.

Our analysis reveals a significant divergence between in-sample and out-of-sample performance. While models demonstrated strong in-sample performance (R²: 0.749–0.812), their predictive power collapsed in out-of-sample testing, often yielding negative R² values. Feature importance analysis shows that price-based features overwhelmingly drive model decisions, accounting for over 60% of importance, while widely used technical indicators such as RSI and Bollinger Bands contribute only 14–15%.

Although integrating technical indicators slightly improved risk-adjusted metrics—yielding Rachev ratios between 0.919 and 0.961—models incorporating these indicators still underperformed a simple buy-and-hold strategy, generating negative returns between -2.4% and -3.9%. These results suggest that traditional technical indicators may be more effective for risk management than for return prediction in HFT settings.

Our findings underscore the importance of adaptive feature selection and regime-specific modeling over reliance on conventional technical indicators. Moreover, the stark contrast between in-sample and out-of-sample performance highlights the necessity of rigorous out-of-sample validation in algorithmic trading research. This study contributes to ongoing discussions on the limitations of technical indicators in HFT and provides insights for both practitioners and researchers aiming to develop more robust predictive models in high-frequency market environments.

Climate change is a significant concern in the contemporary world. It has become a critical issue worldwide in countries, organizations, industries, and financial markets. This study investigates the effect of the relationship between climate risks (CRs) and climate policy uncertainty (CPU) on ESG stocks in both emerging and developed markets. Using the asymmetric TVP-VAR technique, this study analyzes the dynamic nature of these connections. Monthly data on climate policy uncertainty and climate risks were obtained from economicpolicy.com, while ESG leader indexes from the S&P 500 were collected for the period 2013–2023. The findings reveal that climate policy uncertainty (CPU) and physical risks (PRIs) exhibit a negative connectedness with ESG stocks, implying that rising uncertainty and physical climate threats negatively impact the performance of ESG stocks. In contrast, transition risks (TRIs) display a positive connectedness with ESG stocks, suggesting that as economies transition toward low-carbon policies and sustainable practices, ESG investments benefit from these structural changes. This study suggests that policymakers and market participants should take proactive measures to mitigate climate policy uncertainty and address physical climate risks to enhance the stability of the ESG market. This is particularly crucial for emerging markets, where these effects may be more pronounced. By managing climate-related risks effectively, policymakers can foster a more resilient ESG investment environment, ensuring its long-term stability and investor confidence in sustainable markets.

Crowdfunding is increasingly becoming dominant in providing funding for entrepreneurs in emerging economies. Inexperienced entrepreneurs can lead to poorly created campaigns, inadequate marketing strategies, and ineffective communication with potential backers. As a result, projects may fail to reach their funding goals, limiting access to capital for innovative ideas and delaying economic growth. Hence, this study investigates the influence of entrepreneurs' experience on the success of crowdfunding projects in Africa. Based on signalling and human capital theories, it investigates how an experienced entrepreneur signals competence and credibility to potential backers to maximise the probability of crowdfunding success.

Additionally, examine the role of management experience, expertise and knowledge of crowdfunding success. A quantitative research approach and a quantitative analysis of crowdfunding campaigns across various African platforms were employed. Drawing from secondary cross-section data from Kickstarter and Indiegogo, this research uses the probit regression method to analyse and test the hypotheses. The results revealed that entrepreneurs’ experiences of frequently asked questions and the presence of images have positively influenced crowdfunding success. Conversely, targeted amounts and longer duration have negatively affected the crowdfunding success. The findings reveal that entrepreneurs with experience tend to achieve higher crowdfunding success rates. This study contributes to the relative lack of research on crowdfunding in Africa by highlighting the importance of entrepreneurs' experience in crowdfunding success. It provides knowledge for entrepreneurs and investors, focusing on the role that experience plays when managing the complexity of crowdfunding platforms in the African context. The findings are based on signalling and human capital theories. However, the findings cannot be generalised to other developed economies.

This paper contributes to the growing yet limited literature on intra-European remittances, focusing on ten countries: four Nordic countries, three Baltic countries, the United Kingdom, Ireland, and Germany. We find that amount of remittances is positively associated with the volatility of the exchange rate in the residence country vs euro, suggesting that exchange rate risk aversion is one of the drivers of remitting behavior. We use the World Bank estimates of bilateral remittances for 2010–2017 and distinguish naturalized migrants from those holding the citizenship of their country of origin. The positive association between remittances and the volatility of the exchange rate is confirmed by a wide range of panel data methods, including Poisson models with multiple absorbed fixed effects, loglinear mixed effects models, as well as generalized least square estimates for models with heteroskedastic and correlated panels. The control variables include the stock of migrants, PPP-adjusted per capita income in the migrants’ country of residence and country of origin, and time fixed effects. More detailed specifications reveal that the above-mentioned association is positive only where/when at least one-third of the stock of working-age migrants are females, otherwise remittances are negatively associated with the volatility of the exchange rate in the residence country vs euro. This is consistent with an assumption that female migrants try to avoid the exchange rate risk by remitting more from countries (and/or in times) with higher exchange rate volatility, while male migrants are willing to accept this risk waiting for a better exchange rate.

Foreign direct investment (FDI) is a key economic phenomenon and a key driver of economic growth, thereby making its accurate forecasting crucial for policymakers and investors. It further brings capital, technology, and expertise into emerging markets, fostering job creation and innovation. The current study compares four machine learning models—support vector regression (SVR), a Deep Neural Network (DNN), empirical-mode-decomposition-based SVR, and an empirical-mode-decomposition-based DNN—to improve the forecasting accuracy for foreign direct investment using the exchange rate and gross domestic product as the independent variables. The empirical mode decomposition technique is applied to decomposing the series into intrinsic mode functions (IMFs) before feeding it into the machine learning model(s). The models' forecasting performance is evaluated using the mean squared error (MSE), the root mean squared error (RMSE), the mean absolute error (MAE), the mean absolute percentage error (MAPE), the symmetric mean absolute percentage error (SMAPE), and the mean bias deviation (MBD). The results demonstrate that the EMD-based SVR model outperformed all of the other models, achieving the highest accuracy due to its ability to filter noise and capture economic noise. Furthermore, it is shown that decomposition-based hybrid models are effective in financial forecasting, and they provide valuable insights for economic decision-making. Future research could explore other machine learning models and add more macroeconomic variables.

This study explores the diversification potential of cryptocurrencies in traditional investment portfolios and their impact on portfolio performance. Focusing on Bitcoin, Ethereum, and Binance Coin, this research examines their return characteristics, volatility, and correlation with conventional asset classes. A portfolio optimization approach using Minimum Variance and Maximum Sharpe Ratio strategies is applied to assess the benefits of including these digital assets. Historical data from 2019 to 2024 is utilized to compare cryptocurrency-inclusive portfolios with those composed solely of traditional assets such as bonds, gold, and equities.

The findings reveal that integrating cryptocurrencies enhances portfolio returns and improves risk-adjusted performance, with Ethereum and Binance Coin emerging as key return drivers. This study highlights how cryptocurrencies offer unique diversification benefits due to their low correlation with traditional assets, providing investors with new opportunities for optimizing risk and return. Additionally, portfolio optimization results demonstrate that the strategic allocation of digital assets can significantly enhance performance without compromising overall portfolio stability.

By offering empirical evidence on the role of cryptocurrencies in investment strategies, this study contributes to the growing body of research on digital asset integration. The results provide valuable insights for investors and portfolio managers seeking to enhance diversification and capitalize on the evolving financial landscape.

The management of tourism and public catering establishments in Romania is increasingly leveraging financial innovations and emerging technologies to enhance its efficiency and profitability. This study explores the impact of artificial intelligence (AI), machine learning (ML), big data, and cloud computing on financial decision-making, pricing strategies, and customer experience in the hospitality sector. For instance, the AI-powered dynamic pricing models used by Romanian hotels have improved their revenue by up to 20%, while ML-driven demand forecasting helps restaurants reduce their food waste by 15-30%. Algorithmic trading and robo-advisors are also being integrated into financial management strategies, helping optimize investment decisions and resource allocation. Algorithmic trading and robo-advisors are increasingly being used for financial optimization, improving investment returns by an estimated 12-18% annually. Cloud-based financial analytic platforms have contributed to a 10-15% reduction in operational costs, enabling real-time financial monitoring and strategic decision-making. Additionally, blockchain applications in financial reporting and auditing are enhancing transparency and reducing fraud risks. By leveraging computational finance tools, Romanian tourism and hospitality businesses can enhance their financial sustainability, increase their competitiveness, and adapt to the evolving digital economy. This study contributes to the growing body of research on AI-driven financial innovations in the service industry, offering valuable insights for policymakers and business leaders.

Real estate is an asset class that plays a crucial role in home ownership, economic stability, wealth accumulation, and investment portfolio management. Therefore, predicting prices and future market trends is important for home buyers, investors, and policymakers as they help in making informed decisions. Machine learning (ML) has emerged as a useful tool for predictive modeling in financial decisions. The primary objective of this study is to compare and identify the ML algorithms which provide the most accurate predictions for residential housing prices. To achieve this objective, we utilized the Housing Price Index (HPI) from Canada and Australia to analyze performance and influencing factors in this study. Key economic indicators included in the dataset are price-to-income ratio, population growth, interest rates, yield of the 10-year government bond, household real disposable income, and the impact of Covid-19. The data collected span from September 2003 to December 2022, encompassing an analysis of market fluctuations for both markets for approximately two decades. Multiple machine learning algorithms for predictive modeling were used in this study, including K-Nearest Neighbors (KNN), Decision Tree, Random Forest, Bayesian Regression, and Extreme Gradient Boosting (XGBoost). We evaluated model performance and standard error metrics using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R²) to measure accuracy. The results indicated that Bayesian Regression outperformed all other algorithms, followed by XGBoost, for all datasets in terms of accurate prediction. Overall, our study demonstrates the potential of integrating machine learning into real estate analytics and highlights its importance for improving investment strategies and decisions in the residential housing market.