Sun-Yong Choi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Sun-Yong Choi has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Do-Hyeon Kim
Department of Finance and Big Data, Gachon University, Seongnam 13120, Republic of Korea
6 shared publications Dong-Jun Kim
Gachon University
4 shared publications- Dong-Jun Kim
Department of Finance and Big Data, Gachon University, Seongnam 13120, Republic of Korea
3 shared publications - Eunjung Noh
Department of Mathematics, Rollins College, 1000 Holt Ave, Winter Park, FL 32789, USA
3 shared publications - Hui-Sang Kim
Department of Financial Mathematics, Gachon University, Seongnam-si 13120, Republic of Korea
3 shared publications - Ji-Won Kang
Department of Finance and Big Data, Gachon University, Seongnam 13120, Gyeonggi, Republic of Korea
3 shared publications
Latest Publications
- Open Access
Does Climate Policy Uncertainty Shape Carbon-Intensive Stock Networks? Evidence from Connectedness and Entropy
We ask whether climate policy uncertainty (CPU) is associated with how strongly shocks are shared within a daily return network of the 20 largest continuously listed U.S. corporate greenhouse-gas emitters, and compare it with economic policy uncertainty (EPU) and geopolitical risk (GPR). Connectedness is measured with a time-varying parameter vector autoregression (TVP-VAR) and generalized forecast-error variance decompositions, together with the Shannon entropy of each firm’s variance shares; because daily returns are only weakly predictable, these measures essentially describe the time-varying contemporaneous correlation structure of returns. A parallel network of 20 large green-energy-related firms serves as a benchmark. Using data from 2015 to 2025, we find no evidence that CPU is associated with connectedness in either network, in levels or in month-to-month changes, in contrast to much of the literature linking climate policy uncertainty to energy and carbon markets. Economic policy uncertainty, by contrast, is positively associated with connectedness, most robustly in changes. The emitter network is consistently more strongly connected than the green benchmark (average total connectedness of 77% versus 64%), with regulated electric utilities forming its most strongly correlated block, and connectedness in both networks peaks during market-wide stress. A portfolio-level quantile analysis shows higher connectedness in both tails than near the median, although part of this pattern reflects links among the uncertainty indices themselves.
Open AccessDoes Climate Policy Uncertainty Shape Carbon-Intensive Stock Networks? Evidence from Connectedness and Entropy
We ask whether climate policy uncertainty (CPU) is associated with how strongly shocks are shared within a daily return network of the 20 largest continuously listed U.S. corporate greenhouse-gas emitters, and compare it with economic policy uncertainty (EPU) and geopolitical risk (GPR). Connectedness is measured with a time-varying parameter vector autoregression (TVP-VAR) and generalized forecast-error variance decompositions, together with the Shannon entropy of each firm’s variance shares; because daily returns are only weakly predictable, these measures essentially describe the time-varying contemporaneous correlation structure of returns. A parallel network of 20 large green-energy-related firms serves as a benchmark. Using data from 2015 to 2025, we find no evidence that CPU is associated with connectedness in either network, in levels or in month-to-month changes, in contrast to much of the literature linking climate policy uncertainty to energy and carbon markets. Economic policy uncertainty, by contrast, is positively associated with connectedness, most robustly in changes. The emitter network is consistently more strongly connected than the green benchmark (average total connectedness of 77% versus 64%), with regulated electric utilities forming its most strongly correlated block, and connectedness in both networks peaks during market-wide stress. A portfolio-level quantile analysis shows higher connectedness in both tails than near the median, although part of this pattern reflects links among the uncertainty indices themselves.
- 2 June 2026
- Published by Taylor & Francis in Asia Pacific Journal of Tourism Research
- p. 1-20
Understanding the determinants of international bleisure travel satisfaction: a comparative SHAP analysis of inbound bleisure and leisure travelers in South Korea from cross-cultural and activity-based perspectives
1. Bleisure is a form of travel in which travelers integrate leisure activities into business trips, representing a nexus between work and tourism in which individuals switch between the roles of w...
- 2 June 2026
- Published by Taylor & Francis in Asia Pacific Journal of Tourism Research
- p. 1-20
Understanding the determinants of international bleisure travel satisfaction: a comparative SHAP analysis of inbound bleisure and leisure travelers in South Korea from cross-cultural and activity-based perspectives
1. Bleisure is a form of travel in which travelers integrate leisure activities into business trips, representing a nexus between work and tourism in which individuals switch between the roles of w...
- 25 May 2026
- Published by Elsevier in Finance Research Letters
- Vol. 105
Sentiment, fan engagement, and sponsor stock performance: Evidence from alternative data
- 25 May 2026
- Published by Elsevier in Finance Research Letters
- Vol. 105
Sentiment, fan engagement, and sponsor stock performance: Evidence from alternative data
- Open Access
- 23 May 2026
- Published by MDPI AG in Fractal and Fractional
- Vol. 10 (6) , 353
Comparative Analysis of Cryptocurrency Market Efficiency and Local Features Using MF-DFA and DCC-GARCH
This study investigates time-varying market efficiency and cross-market correlations in cryptocurrency markets across South Korea, the United States, and Japan. Using rolling-window multifractal detrended fluctuation analysis (MF-DFA) and dynamic conditional correlation–generalized autoregressive conditional heteroskedasticity (DCC-GARCH), we analyze 11 cryptocurrency–fiat pairs—Bitcoin (BTC), Ethereum (ETH), Ripple (XRP), and Bitcoin Cash (BCH) denominated in Korean Won (KRW), US Dollar (USD), and Japanese Yen (JPY)—from January 2018 to September 2025. MF-DFA results confirm persistent multifractality and significant time-variation in market efficiency across all markets, consistent with the Adaptive Market Hypothesis (AMH). DCC-GARCH estimates reveal a structural divergence between return integration and efficiency correlations: return-based correlations for same-asset cross-fiat pairs are exceptionally high (mean dynamic conditional correlation of approximately 0.96–0.98), whereas efficiency-based correlations are far more heterogeneous, with cross-asset pairs approaching near-zero synchronization. We interpret the Kimchi Premium as a product of institutional frictions that impede price-level arbitrage while leaving volatility transmission largely unaffected. These findings suggest that cryptocurrency market integration is multidimensional—globally synchronized in risk dynamics, yet locally segmented in the structural quality of information processing.
- 23 May 2026
- Published by MDPI AG in Fractal and Fractional
- Vol. 10 (6) , 353
Open AccessComparative Analysis of Cryptocurrency Market Efficiency and Local Features Using MF-DFA and DCC-GARCH
This study investigates time-varying market efficiency and cross-market correlations in cryptocurrency markets across South Korea, the United States, and Japan. Using rolling-window multifractal detrended fluctuation analysis (MF-DFA) and dynamic conditional correlation–generalized autoregressive conditional heteroskedasticity (DCC-GARCH), we analyze 11 cryptocurrency–fiat pairs—Bitcoin (BTC), Ethereum (ETH), Ripple (XRP), and Bitcoin Cash (BCH) denominated in Korean Won (KRW), US Dollar (USD), and Japanese Yen (JPY)—from January 2018 to September 2025. MF-DFA results confirm persistent multifractality and significant time-variation in market efficiency across all markets, consistent with the Adaptive Market Hypothesis (AMH). DCC-GARCH estimates reveal a structural divergence between return integration and efficiency correlations: return-based correlations for same-asset cross-fiat pairs are exceptionally high (mean dynamic conditional correlation of approximately 0.96–0.98), whereas efficiency-based correlations are far more heterogeneous, with cross-asset pairs approaching near-zero synchronization. We interpret the Kimchi Premium as a product of institutional frictions that impede price-level arbitrage while leaving volatility transmission largely unaffected. These findings suggest that cryptocurrency market integration is multidimensional—globally synchronized in risk dynamics, yet locally segmented in the structural quality of information processing.
- 15 April 2026
- Published by Elsevier in Finance Research Letters
- Vol. 100
Liquidity depth and information efficiency: High-frequency evidence from leading U.S. technology stocks
- 15 April 2026
- Published by Elsevier in Finance Research Letters
- Vol. 100
Liquidity depth and information efficiency: High-frequency evidence from leading U.S. technology stocks
- Open Access
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in AIMS Mathematics
- Vol. 11 (1) , 977-1020
A hybrid transformer framework integrating sentiment and dynamic market structure for stock price movement forecasting
Traditional financial forecasting models are inherently limited as their sole reliance on price data causes them to overlook critical information such as market sentiment and dynamic inter-market interactions. To address this shortcoming, this study proposes a novel hybrid transformer model that integrates heterogeneous data sources. Specifically, our framework combines investor sentiment extracted from news headlines using FinBERT and dynamic changes in market structure analyzed with a TVP-VAR model, along with traditional log-return data. In forecasting experiments conducted on four major global stock indices (S&P 500, FTSE 100, CSI 300, and Nikkei 225), the proposed model consistently demonstrated superior performance compared to both single-data-source models and traditional benchmarks. We found that the inclusion of dynamic market structure information, derived from the TVP-VAR model, as a predictive variable was a decisive factor in improving forecasting accuracy. This research empirically validates that a multi-modal approach combining heterogeneous data can significantly enhance the precision of financial market forecasting.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in AIMS Mathematics
- Vol. 11 (1) , 977-1020
Open AccessA hybrid transformer framework integrating sentiment and dynamic market structure for stock price movement forecasting
Traditional financial forecasting models are inherently limited as their sole reliance on price data causes them to overlook critical information such as market sentiment and dynamic inter-market interactions. To address this shortcoming, this study proposes a novel hybrid transformer model that integrates heterogeneous data sources. Specifically, our framework combines investor sentiment extracted from news headlines using FinBERT and dynamic changes in market structure analyzed with a TVP-VAR model, along with traditional log-return data. In forecasting experiments conducted on four major global stock indices (S&P 500, FTSE 100, CSI 300, and Nikkei 225), the proposed model consistently demonstrated superior performance compared to both single-data-source models and traditional benchmarks. We found that the inclusion of dynamic market structure information, derived from the TVP-VAR model, as a predictive variable was a decisive factor in improving forecasting accuracy. This research empirically validates that a multi-modal approach combining heterogeneous data can significantly enhance the precision of financial market forecasting.
- Open Access
- 22 October 2025
- Published by MDPI AG in Electronics
- Vol. 14 (21) , 4136
A Hybrid Frequency Decomposition–CNN–Transformer Model for Predicting Dynamic Cryptocurrency Correlations
This study proposes a hybrid model that integrates Wavelet frequency decomposition, convolutional neural networks (CNNs), and Transformers to predict correlation structures among eight major cryptocurrencies. The Wavelet module decomposes asset time series into short-, medium-, and long-term components, enabling multi-scale trend analysis. CNNs capture localized correlation patterns across frequency bands, while the Transformer models long-term temporal dependencies and global relationships. Ablation studies with three baselines (Wavelet–CNN, Wavelet–Transformer, and CNN–Transformer) confirm that the proposed Wavelet–CNN–Transformer (WCT) consistently outperforms all alternatives across regression metrics (MSE, MAE, RMSE) and matrix similarity measures (Cosine Similarity and Frobenius Norm). The performance gap with the Wavelet–Transformer highlights CNN’s critical role in processing frequency-decomposed features, and WCT demonstrates stable accuracy even during periods of high market volatility. By improving correlation forecasts, the model enhances portfolio diversification and enables more effective risk-hedging strategies than volatility-based approaches. Moreover, it is capable of capturing the impact of major events such as policy announcements, geopolitical conflicts, and corporate earnings releases on market networks. This capability provides a powerful framework for monitoring structural transformations that are often overlooked by traditional price prediction models.
- 22 October 2025
- Published by MDPI AG in Electronics
- Vol. 14 (21) , 4136
Open AccessA Hybrid Frequency Decomposition–CNN–Transformer Model for Predicting Dynamic Cryptocurrency Correlations
This study proposes a hybrid model that integrates Wavelet frequency decomposition, convolutional neural networks (CNNs), and Transformers to predict correlation structures among eight major cryptocurrencies. The Wavelet module decomposes asset time series into short-, medium-, and long-term components, enabling multi-scale trend analysis. CNNs capture localized correlation patterns across frequency bands, while the Transformer models long-term temporal dependencies and global relationships. Ablation studies with three baselines (Wavelet–CNN, Wavelet–Transformer, and CNN–Transformer) confirm that the proposed Wavelet–CNN–Transformer (WCT) consistently outperforms all alternatives across regression metrics (MSE, MAE, RMSE) and matrix similarity measures (Cosine Similarity and Frobenius Norm). The performance gap with the Wavelet–Transformer highlights CNN’s critical role in processing frequency-decomposed features, and WCT demonstrates stable accuracy even during periods of high market volatility. By improving correlation forecasts, the model enhances portfolio diversification and enables more effective risk-hedging strategies than volatility-based approaches. Moreover, it is capable of capturing the impact of major events such as policy announcements, geopolitical conflicts, and corporate earnings releases on market networks. This capability provides a powerful framework for monitoring structural transformations that are often overlooked by traditional price prediction models.
- Open Access
- 21 July 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (7) , e0325106
Baltic dry index forecast using financial market data: Machine learning methods and SHAP explanations
The Baltic Dry Index (BDI) is a critical benchmark for assessing freight rates and chartering activity in the global shipping market. This study forecasts the BDI using diverse financial data, including commodities, currencies, stock markets, and volatility indices. Unlike previous research, our approach integrates financial indicators specific to major marine trading regions—the U.S., EU, and Hong Kong. We employ advanced machine learning methods, such as Extremely Randomized Trees, Categorical Boosting (CatBoost), and Random Forest, to achieve superior forecasting accuracy. Additionally, we utilize the Shapley Additive Explanations (SHAP) framework to analyze the contributions of financial features to BDI predictions. Key findings reveal that the S&P 500 index is the most influential factor, followed by significant contributions from iron ore and coal commodity indices and the dollar index, underscoring the interplay between the U.S. economy and the BDI. By integrating SHAP explanations, this study not only predicts market trends but also uncovers the economic drivers shaping the BDI. Practically, it supports the stability of the global shipping industry by enabling more informed decision-making for stakeholders. Academically, it introduces overlooked economic factors in BDI prediction, offering valuable insights and directions for future research.
- 21 July 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (7) , e0325106
Open AccessBaltic dry index forecast using financial market data: Machine learning methods and SHAP explanations
The Baltic Dry Index (BDI) is a critical benchmark for assessing freight rates and chartering activity in the global shipping market. This study forecasts the BDI using diverse financial data, including commodities, currencies, stock markets, and volatility indices. Unlike previous research, our approach integrates financial indicators specific to major marine trading regions—the U.S., EU, and Hong Kong. We employ advanced machine learning methods, such as Extremely Randomized Trees, Categorical Boosting (CatBoost), and Random Forest, to achieve superior forecasting accuracy. Additionally, we utilize the Shapley Additive Explanations (SHAP) framework to analyze the contributions of financial features to BDI predictions. Key findings reveal that the S&P 500 index is the most influential factor, followed by significant contributions from iron ore and coal commodity indices and the dollar index, underscoring the interplay between the U.S. economy and the BDI. By integrating SHAP explanations, this study not only predicts market trends but also uncovers the economic drivers shaping the BDI. Practically, it supports the stability of the global shipping industry by enabling more informed decision-making for stakeholders. Academically, it introduces overlooked economic factors in BDI prediction, offering valuable insights and directions for future research.