Probability of stock price crash and the expected return of stock under sanctions
Volume 8, Issue 1, 2024, Pages 131-145
https://doi.org/10.61186/ijf.2024.397009.1411
Najibeh Najafi Kangarloui, Farkhondeh Jebel Ameli, Mohsen Mehrara, Mir Behnam Fatehi
Abstract Sanctions increase managers' motivations to manage profits and the accumulation of bad news due to their negative impact on corporate profitability and cash flow. According to Jin and Myers (2006), this issue increases companies' stock price crash risk. The higher probability of stock price crashes indicates a stock price overvaluation. As a result, the expected return on such shares will be low. The current paper used the softmax model to calculate the probability of stock price crashes and the expected return calculated by the Fama-French three-factor model. The sample used in this paper includes 80 import- and export-oriented exchange companies from 2008-2021. The results of this paper indicate a positive and significant relationship between the sanctions variables and the probability of stock crashes. So, sanctions cause an increase in the accumulation of bad news and information asymmetry between managers and investors. The second part of the paper's results indicates a negative relationship between the probability of stock crashes and the expected return on the stocks in the Iranian capital market. Therefore, investors have relatively good analytical skills in the Iranian capital market due to its shallow depth and infrastructure problems. The results of this paper can be used in portfolio management to select stocks with a lower probability of crash and higher return.
Identifying and Prioritizing the Factors Affecting Enterprise Risk Management Implementation
Volume 9, Issue 1, Winter 2025, Pages 134-161
https://doi.org/10.61186/ijf.2024.448623.1463
Kourosh Zarinjooei, Hasan Ghodrati, Hossein Panahian, Hossein Jabbari
Abstract Enterprise risk management (ERM) represents a new paradigm that supports organizations in identifying, evaluating, and managing risks. Several factors encourage different organizations, especially banks, to design and apply ERM; among them, the possibility of financial problems and related costs, financial performance decline, growth opportunities, and independence of the board of directors can be pointed out. In addition, applying a suitable risk management strategy is a competitive advantage for supporting companies. Following the previous comprehensive studies, the present study was done by integrating a meta-composition approach and multivariate fuzzy network analysis. In this research, to identify effective factors on ERM based on empirical evidence and selection of studies, description and classification of selected articles, analysis of the content of selected research articles, and finally, the importance and refinement of the identified factors based on the Delphi expert opinion polling technique, and multivariate fuzzy network analysis was discussed. The study aimed to identify and rank the factors affecting the effectiveness of ERM of the firms accepted in selected banks in Iran. While reviewing studies, a semi-structured interview was conducted using an Exploratory-Descriptive Qualitative (EDQ) research design to determine the factors that affect ERM effectiveness. The interviewed experts were comprised of 20 university professors, CEOs, financial analysts of investment and brokerage companies, and senior auditors with accounting and management education. The literature review and the results of the interviews indicate five main themes that classify the factors affecting ERM effectiveness in firms. In this study, key factors were identified, and then the fuzzy Delphi technique was employed to rank and find the weight of the factors. The results showed that corporate governance, financial indicators, environmental indicators, company characteristics, and management indicators effectively enhance ERM's effectiveness. Accordingly, capital market analysts and investment companies should take a broader perspective and make decisions based on companies' financial risks instead of paying attention to companies' profitability and stock price changes. The results demonstrated that five factors determine ERM effectiveness: 1. corporate governance indicators (monitoring the board of directors and ownership structure), 2. Financial indicators (return on assets (ROA), earnings volatility, merger and acquisition (M&A) activities, financial deficiency, and capital opacity), 3. Environmental indicators (performance excellence, industry competition, audit firm credibility, environmental uncertainty, and industry), 4. Firm characteristics (financial leverage, firm size, and growth opportunities), and 5. Management indicators (management career and business diversity).
Choosing Influential Dimension of Mindfulness of Auditors' Professional Judgments Based on Social Pressure: An Analysis Based on Rough Theory
Volume 6, Issue 2, Spring 2022, Pages 135-166
https://doi.org/10.30699/ijf.2021.279873.1214
Seyed Mohammad Soleymani, Farhad Dehdar, Mohammadreza Abdoli
Abstract The growth and complexity of society justify the need for relevant economic information, information systems, and information-generating processes, and the need for auditing as part of the information reporting process increases. This has affected the professional functions of auditors. The purpose of this research is to choose the influential dimension of mindfulness of auditors' professional judgments based on social pressure analysis based on Rough Theory. The methodology of this research is mixed and it has been used by Meta-synthesis, Delphi and Rough Theory. The target population was the qualitative, similar research and academic experts in the field of accounting. However, the target population in a small number of 19 audit partners had more than 5 years of work experience, which is acceptable from the statistical population due to the requirement of Ruff theory analysis. In this study, based on the Meta-synthesis analysis of selected researches, 4 propositions of pressures based on social compliance and 3 components of auditors' professional judgment were determined. The results in this section show that the most effective proposition of social compliance pressures was the market pressure proposition, which affects the inferential consciousness of auditors in their professional judgment and violates the auditors' mental functions such as skepticism and objectivity.
Accounting Modeling for Startups in the Financial Business Ecosystem
Volume 9, Issue 3, 2025, Pages 135-165
https://doi.org/10.66224/ijf.2025.513542.1509
Touraj Bahrami, Saeid Jabbarzadeh Kangarluei, Ali Ashtab, Masoud Ahmadi Mansourabad
Abstract This research aims to model the accounting of startup companies within the financial business ecosystem.This research is philosophically based on the interpretivist paradigm and was conducted with an inductive approach. It is also applied - developmental in terms of purpose and descriptive in terms of method and data collection timeframe. A nested research design was used to achieve the research objective. The research participants included theoretical experts (professors of financial management and accounting) and empirical experts (managers of startup companies). Theoretical sampling was used until theoretical saturation was reached, and eventually, 24 individuals participated in the study. Data collection tools included semi-structured interviews and a questionnaire based on a decision matrix. The validity of the interview was confirmed based on four criteria: credibility, transferability, confirmability, and dependability. The reliability of the qualitative section was estimated to be favorable by calculating Holst's coefficient at 0.817 and Cohen's Kappa coefficient at 0.706. Data analysis was performed using grounded theory in the qualitative section and the SWARA method in the quantitative section. Based on the research model, it was determined that causal conditions (technical factors, managerial factors, governmental factors, and accounting factors) influence the core phenomenon (startup accounting). The core phenomenon, contextual conditions (financial business ecosystem infrastructure and financial business ecosystem financial resources), and intervening conditions (financial business ecosystem regulations) influence strategies and actions (technological strategy and financial strategy). This research provides a comprehensive model for understanding startup accounting within the financial business ecosystem. The findings highlight the critical influence of technical, managerial, governmental, and accounting-related factors on core accounting practices. Furthermore, the interplay between these practices, the broader ecosystem infrastructure, financial resources, and the regulatory framework shapes startups' strategic decisions (both technological and financial). These strategic choices, in turn, directly affect both the financial and non-financial performance outcomes of these nascent firms. The model underscores the need for a nuanced approach to startup accounting that takes into account the specific context of the financial business ecosystem.
Challenges of Utilizing Crypto-currencies in Capital Markets: An Islamic Jurisprudence Approach
Volume 8, Issue 4, 2024, Pages 136-151
https://doi.org/10.30699/ijf.2025.433073.1454
Reza Mirzakhani, Farhad Morsali Pavarsi
Abstract Money as a medium of exchange signifies anything widely accepted as a means of payment. There are two types of money: commodity money, any money deriving its price from its own intrinsic and inherent value, such as gold or silver, and fiat money, the typical money that does not have intrinsic value but the backing of a government. The concept of digitalization of money was aimed at the enhancement of the function, facilitation, and utilization of money as a means of medium. The first cryptocurrency was created by Satoshi Nakamoto, the pseudonym for an anonymous computer programmer or group of programmers, on January 3, 2009. Muslim countries have been engaged and involved in approaching cryptocurrencies in accordance with Shariah and Islamic principles as new and novel types of money. Cryptocurrencies may be beneficial instruments in capital markets, such as acting as a financial instrument, a medium for clearing and settlement, or an underlying asset in derivative contracts. However, some certain aspects of crypto-currencies can be challenged by Islamic finance and Shariah when it comes to their functions in financial markets, especially, capital markets, such as their cryptic and ambiguous nature rendering Gharar in transactions or difficulty in valuating and accurate pricing due to their volatility and ambiguity or the technology behind these assets. Utilizing cryptocurrencies in financial systems may also cause systemic disorder, which may damage financial plumbing and systems, leading to losses for investors and market practitioners and creating profits for some with no effort (Akl bil batil). The challenges of cryptocurrencies, including security problems in the general aspect, and accurate valuation based on the nuances between their intrinsic values and nominal values, among others, may make the process of institutionalization of cryptocurrencies in the community, more specifically in Islamic capital markets, hard.
Risk prediction of investment funds in member countries of the Federation of European and Asian Stock Exchanges - Machine Learning Approaches
Volume 9, Issue 4, 2025, Pages 140-188
https://doi.org/10.66224/ijf.2025.522399.1527
Nashmil Esmaily, Parviz Piri, Ali Ashtab, Mehdi Heydari, Akbar Zavari Rezaei,
Abstract The main objective of this study is to compare the predictive accuracy of machine learning models, particularly Random Forest and Artificial Neural Networks, with classical statistical methods (such as Logistic Regression and Linear Discriminant Analysis) in forecasting the risk of Exchange-Traded Funds (ETFs) in member countries of the Federation of European and Asian Stock Exchanges. Furthermore, the study aims to identify the key performance and fundamental variables impacting the risk of these funds. This research adopts a quantitative approach based on secondary data analysis. Data were collected for the years 2015-2023 from the databases of the Federation of European and Asian Stock Exchanges and the Tehran Stock Exchange. After preprocessing, risk prediction models, including Random Forest, Artificial Neural Networks, Logistic Regression, and Linear Discriminant Analysis, were developed and validated for each country using unified evaluation metrics (such as accuracy and AUC). The statistical significance of differences in model performance was tested using non-parametric Mann-Whitney U tests, given the non-normal distribution of accuracy across countries. Sensitivity analysis was then conducted on the two superior machine learning models to determine the impact of independent variables (both performance indicators, such as Jensen's alpha and market return, and fundamental attributes, such as fund size and manager expertise) across different markets. Empirical results indicate that, across most countries and after harmonizing time and geographical dimensions, machine learning models, specifically Random Forest and Artificial Neural Networks, outperform classical statistical approaches in predicting ETF risk, with statistically significantly higher accuracy and AUC values (p<0.05 in Mann-Whitney U tests). The robustness of these findings is confirmed after controlling for heterogeneity among countries. Sensitivity analyses further reveal that both performance variables (e.g., Jensen's alpha, market return) and fundamental factors (e.g., fund size, manager expertise) have a significant impact on risk outcomes within these models. At the same time, machine learning methods exhibit a stronger ability to identify and quantify the importance of these variables compared to classical methods. The results highlight the practical advantage of adopting machine learning techniques for risk assessment and management in diverse international financial markets. Overall, the findings of this study reveal that employing machine learning models—especially Random Forest and Artificial Neural Networks—significantly improves the accuracy of ETF risk prediction and enables a more comprehensive identification of key risk factors compared to classical statistical approaches. These models demonstrate superior flexibility and the ability to capture complex, multidimensional data patterns, making them highly advantageous tools for financial risk management. The results suggest that integrating advanced machine learning techniques at both regional and international levels can enhance the responsiveness of investment systems to market changes, providing fund managers and investors with a more solid, data-driven basis for decision-making.
Identification of the Factors Affecting Auditors’ Conflict of Interests Using Fuzzy Delphi Method
Volume 6, Issue 1, Winter 2022, Pages 142-168
https://doi.org/10.30699/ijf.2021.250031.1159
Javad Ghaznavi Doozandeh, Mansour Garkaz, Ali Khozein, Alireza Maetoofi
Abstract The present study aimed to identify the most effective causes of conflict of interest by examination of accounting literature and expert consensus. Understanding these factors, using cognitive psychology theories, can lead to a model for reducing conflict of interests. The dignity of the audit profession depends on fair and proper professional judgment by auditors, and achieving this requires identification and controlling of the key factors affecting judgment and decision-making. When auditors intentionally or unintentionally accredit financial statements in line with the opinion of their employers, public interests and the auditing profession are at serious risk. Several factors that can categorize into seven categories of structure, community, culture, environment, personality, audit firm characteristic, and ethics and behavior are rooted in a conflict of interests. However, no comprehensive research examining all the above factors and identifying the most effective ones has been done so far. By reviewing the research literature, major and minor factors were identified in domestic and foreign sources. Ten expert auditors were selected by the snowball method and interviewed. The considered major and minor factors were selected from among the introduced factors, and a questionnaire was sent to the experts using the Fuzzy Delphi (Screening) method. The results of the above statistical analysis identified eighteen of the most prominent sub-criteria of the factors affecting conflict of interests and identified structural factors the highest rank in this classification, which was agreed by the experts.
Credit Scoring Active Telegram Channels Offering Stock Signals
Volume 6, Issue 3, Spring 2022, Pages 142-164
https://doi.org/10.30699/ijf.2022.309877.1279
Babak Sohrabi, Ahmad Khalili Jafarabad, Saba Orfi
Abstract The impact of personal judgment on the assessment of an individual’s financial situation has been drastically reduced through the development of credit scoring. The systems are capable of deciding based on an applicant’s total score which is a combination of several factors and indicators. Over the past few decades, credit scoring has been considered an essential tool for evaluation in various institutions and has also been able to transform the industry as a whole. Most of the research conducted in the field has taken into account traditional credit scoring, but considering the ever-evolving technological world that we live in and the increasing emergence of new social media networks, such research has now become obsolete. Such technological advancements have not only paved the way for far more sophisticated credit scoring systems but also essentially rendered the previous generations useless. It should be noted that credit scoring and its features have widely been discussed across the globe but, considering the various aspects and models that have to be taken into account, no one best method has been designed or suggested for it so far. This study shows that social media channels tend to perform relatively well in predicting stock market trends when the overall index is growing positively. The research also illustrates that a higher number of days of activity and a large number of signals released do not necessarily mean that the channels can or have credited their offered stock return on a one-month time frame. The methodology used is "CRISP-DM," which consists of six steps. The main variables include social and financial variables that are examined for six months. In the research, we seek to identify, analyze and categorize active telegram channels in stock signals using the data mining model and the RFM method. The k-means algorithm is selected for this category. Then, in each cluster, the importance of social variables and the performance of the channels are extracted by the EXTRATREECLASSIFIER algorithm, and channel performance is measured by considering the changes in the total index.
Evaluation and comparison net assets value of joint investment funds using support machine models versus statistical models - A case study from FEAS member countries
Volume 7, Issue 4, 2023, Pages 142-160
https://doi.org/10.61186/ijf.2023.376186.1389
Leila Nateghian, Saeid Jabbarzadeh Kangarlouei, Jamal Bahri Sales, Parviz Piri
Abstract Today, choosing the suitable model for determining the portfolio of investment in financial assets is one of the critical issues of the attention of analysts and capital market activists, and investing in a portfolio consisting of mutual investment funds is the same. With this statement, the purpose of the article is to evaluate and compare the net assets value (return) of the Federation of Asian and European Stock Exchanges (FEAS) member countries by using support machine models in comparison with statistical models. The statistical and sample population included the data of 39 selected traded funds and FEAS members from 12 selected countries (including Iran) between 2014 and 2021.
The data related to the mentioned funds were classified and analyzed using spss-modeler, rapid miner, and Weka software. They were tested with 24 support machine methods and 11 statistical methods, and the results showed that the prediction accuracy of statistical models is lower than that of support machine models. The Mann-Whitney test was used to determine the significance of this difference. Also, the results show that at the 95% confidence level, it can be claimed that the prediction accuracy of machine learning models is higher than statistical models. The average rating of machine learning models was (20.86) much higher than statistical models (10.85).
A Neurofinance-Based Model for Developing Public Investor Trust
Volume 10, Issue 1, 2026, Pages 145-172
https://doi.org/10.66224/ijf.2026.562992.1560
Habib Niroomand, Zohreh Khajehsaeed, Asgar Pak Maram
Abstract Iran’s capital market has witnessed substantial transformation over recent decades. However, the sharp downturn of the stock market in 2020 represented a critical juncture, extending beyond retail investors’ financial losses and culminating in a widespread public trust crisis. Addressing this issue, the present study proposes a neurofinance-based model aimed at strengthening public investor trust in Iran’s capital market. The model was developed through a mixed-methods design, integrating qualitative grounded theory exploration with quantitative validation via structural equation modeling (SEM). In the qualitative phase, data were collected in 2025 through semi-structured interviews with 17 capital market experts and analyzed using a three-stage coding procedure comprising: open, axial, and selective coding. In the quantitative phase, the proposed theoretical model was empirically tested using SEM on data obtained from 87 investors in the capital market. The qualitative findings reveal that the development of trust is shaped by causal conditions (e.g., information transparency and emotional responses), contextual conditions (e.g., economic stability and social capital), and intervening conditions (e.g., media, education, and supportive institutions). The quantitative results confirm that all model paths are statistically significant and that the model demonstrates an acceptable level of fit. Accordingly, strategies such as enhancing transparency, empowering retail investors, and promoting financial literacy were proposed, generating outcomes at the individual, market, and macro levels. The novelty of this research lies in integrating a neurofinance perspective with a mixed-methods approach to develop a context-specific model aligned with Iran’s institutional and cultural environment.
Exchange Rate Movements and Monetary Policies: Which Has Greater Influence on Petroleum
Volume 5, Issue 1, Winter 2021, Pages 147-172
https://doi.org/10.30699/ijf.2021.125534
Shahram Molavi Bisetoni, Kiamars Fathi Hafshejani, Aboutorab Alirezaei, Ghanbar Abbaspour Esfadan
Abstract The primary purpose of this study is twofold: Firstly, using the Markov Regime Switching model throughout December 2008 to February 2020, it investigates and compares the nonlinear impacts of exchange rate movements and monetary policies on Petroleum Stock Index, PSI, in Iran. Accordingly, some control variables, such as OPEC oil price, inflation rate, and international sanctions, have also been used to model these relationships more accurately. Secondly, it is an empirical attempt to trace the historical changes in the PSI behavior through distinguishing the precise regime numbers, and the relationships between the exogenous variables and the PSI. Our results confirm that the effects of both exchange rate movements and monetary policies on the petroleum stock market return are direct and significant. More interestingly, the more we move from regime one to regime three, the greater the effects of the research variables on the index, except for the impact of OPEC oil prices. Our empirical findings further suggest as the effects of sanctions intensify, the influences of monetary policy and exchange rate movements would have a more significant impact on the petroleum stock index returns.
Causal Alpha: Identifying Structural Drivers of Excess Returns in the US Equity Market Using Causal Machine Learning
Volume 10, Issue 2, 2026, Pages 158-185
https://doi.org/10.66224/ijf.2026.593966.1593
Moslem Nilchi, Mahdi Zerang
Abstract This study investigates the causal relationship between uncertainty shocks and factor premia in the US equity market. While traditional asset pricing models rely on correlation-based factor structures, this research argues that identifying true alpha necessitates a causal framework. Employing a novel combination of Double Machine Learning (DML) and Causal Forest algorithms on Kenneth French's five-factor data and VIX-based uncertainty shocks from 2001 to 2025, robust evidence is provided demonstrating that uncertainty shocks exert significant causal effects on factor returns. The findings reveal that the market factor (MKT-RF) and the size factor (SMB) exhibit strong negative causal responses to uncertainty shocks. In contrast, the profitability factor (RMW) demonstrates a positive and significant effect. The value (HML) and investment (CMA) factors, however, show no statistically detectable causal effect. Substantial heterogeneity in these effects is documented across high- and low-volatility regimes, with the market's negative response being substantially larger in high-uncertainty periods. From an economic significance perspective, a long-only trading strategy based on causal signals delivers a competitive Sharpe ratio of 0.437, with a substantially lower maximum drawdown (-32.5%) compared to the market portfolio (-51.4%). The results survive rigorous robustness checks and statistical tests, providing evidence consistent with a causal effect of uncertainty shocks on factor premia. This study contributes to the growing literature on causal machine learning in finance and offers practical implications for risk management and portfolio construction.
Designing a Causal Model for Multi-Criteria Decision-Making in Financial Risk Analysis and Financing of IT-Based Startup Companies (BWM-DEMATEL) approach
Volume 9, Issue 1, Winter 2025, Pages 162-198
https://doi.org/10.61186/ijf.2025.482809.1492
Mohammad Mostafa Bod, Reza Raei
Abstract This research uses a fuzzy Delphi approach to identify the dimensions and components of investor risk and financing for IT-based startup companies. The statistical population for this study consisted of 30 experts and university professors familiar with the research concepts surveyed to select the dimensions and components identified from the literature and prior research. The results from the fuzzy Delphi method showed that the financing and investor risk dimensions were selected in 9 dimensions and 29 components. The weighting results for the research dimensions and components using the Best-Worst Method (BWM) prioritized each. The weighting results indicate that the industry status ranked first, followed by scientific factors and other components, ranked third to ninth in the LINGO software. Additionally, the intensity of relationships among the research dimensions was assessed using the Decision-Making Trial and Evaluation Laboratory (DEMATEL) technique. The analysis of the intensity of relationships among the dimensions shows that the government factors dimension had the highest numerical value based on the row sum, making it the most influential dimension among those examined in financing startup companies using the DEMATEL technique. Conversely, based on the D-R analysis, the geographic factors dimension received the lowest value and was recognized as the most affected dimension. Using the fuzzy Delphi method, this research has identified specific and novel dimensions and components, such as governmental, geographical, and scientific factors, which have been less addressed in the existing literature.
Exploring the Role of Artificial Intelligence in Corporate Financial Asset Allocation: Evidence from the Tehran Stock Exchange
Volume 9, Issue 3, 2025, Pages 166-186
https://doi.org/10.66224/ijf.2025.232445
Moslem Nilchi, Mohammad Ali Heidarihaei
Abstract Although previous research has examined the application of artificial intelligence (AI) across various areas of finance, there remains limited empirical evidence regarding its impact on corporate financial asset allocation. This gap is particularly evident when considering the organisational capabilities that enable firms to utilise AI technologies effectively. In the rapidly evolving technological landscape, artificial intelligence (AI) has emerged as a pivotal force driving innovation and transformation within corporate financial management. By embedding AI into organisational processes, companies have fundamentally reshaped their financial decision-making frameworks. However, the exact mechanisms through which AI adoption shapes the allocation of financial assets are still not fully understood. This study examines how artificial intelligence (AI) technologies influence the allocation of financial assets within corporations, with a specific focus on the moderating influence of three dynamic organisational capabilities: absorptive capacity, innovation capability, and adaptability. Based on panel data collected from companies listed on the Tehran Stock Exchange between 2020 and 2024, AI adoption is measured through textual analysis of management commentary reports obtained from the Codal system. The dependent variable comprises a set of financial ratios, including the proportion of financial assets relative to a firm’s total assets. The analysis employs multiple regression models with interaction terms to test the proposed hypotheses. Findings indicate that the adoption of AI substantially enhances the effectiveness of distributing financial assets. Moreover, absorptive capacity and innovation capability strengthen the association between AI adoption and the allocation of financial assets within firms' performance, while adaptability shows no statistically significant moderating effect. These results highlight the importance of both technological infrastructure and internal capabilities for leveraging advanced technologies to their fullest potential. This research not only enriches the academic discourse with fresh empirical insights but also offers valuable implications for financial managers, capital market regulators, and policymakers engaged in organisational digital transformation strategies.
Analysis of the characteristics affecting the trading risk of listed companies' stocks: A hybrid spatial artificial intelligence approach
Volume 10, Issue 1, 2026, Pages 173-237
https://doi.org/10.66224/ijf.2026.569343.1569
Javad Zolfaghary Tabesh, Babak Jamshidinavid, Mehrdad Ghanbary, Afshin Baghfalaki
Abstract This research identifies the determinants of trading risk (conditional variance) in the Iranian stock market over 15 years (2008-2023) using a novel hybrid approach combining spatial econometrics and machine learning algorithms. The main objective is to evaluate the superiority of hybrid models over traditional methods and to identify the roles of macroeconomic, geopolitical, behavioral factors, and firm characteristics in shaping systematic risk. The sample includes 172 companies listed on the Tehran Stock Exchange with 30,960 monthly observations and 33 explanatory variables. The methodology was implemented in three stages: First, GARCH and EGARCH models were employed to extract conditional variance and confirm the leverage effect. Second, the Spatial Durbin Error Model (SDEM) was used to decompose direct, spatial spillover, and total effects of variables while controlling for cross-sectional dependence (Pesaran CD statistic = 87.45***) and spatial autocorrelation (Moran's I = 0.4567***). Third, machine learning algorithms, including Linear Regression, SVM, Random Forest, XGBoost, LSTM, and Transformer, were applied independently and in combination with SDEM outputs. The results demonstrated a clear performance hierarchy: Linear Regression (R² = 0.4123, RMSE = 0.0987), SVM (R² = 0.5987), Random Forest (R² = 0.6789), XGBoost as the best standalone model (R² = 0.7456, RMSE = 0.0534), and Ensemble (R² = 0.7523). Hybrid models showed significant superiority: SDEM + XGBoost (R² = 0.7823, RMSE = 0.0471; 11.80% error reduction compared to standalone XGBoost and 52.3% improvement over Linear Regression), and SDEM + Ensemble (R² = 0.7867, RMSE = 0.0467) achieved optimal performance. Time-series cross-validation (average test RMSE = 0.0492) and the Diebold-Mariano test (DM = 3.456*** against XGBoost) confirmed statistical superiority. From a substantive perspective, the exchange rate with a total effect of 0.2443*** and SHAP contribution of 18.34% was identified as the most important systematic risk factor, followed by sanction intensity (total effect = 0.1274***, SHAP = 14.23%), Altman Z-score (SHAP = 15.67%), total stock index (total effect = -0.1801***, SHAP = 12.89%), and investor sentiment (total effect = 0.1001***, SHAP = 11.45%). The findings demonstrate that hybrid spatial econometrics and machine learning models improve prediction accuracy by 12-15% through extracting complementary information. Geopolitical and behavioral factors, in addition to traditional macroeconomic variables, are systematically important. Spatial spillovers constitute 15-25% of total effects, which are ignored in traditional models. This research shows that the frontier of financial risk modeling lies in the synergistic integration of economic theory and machine learning.