Department of Finance · Da-Yeh University · Changhua, Taiwan
Modelling how financial risk actually behaves.
Simulated conditional volatility · GARCH(1,1) · illustrative
We build statistical models of volatility, dependence, jumps and tail loss — then test whether they survive contact with real market data.
Our methods sit at the intersection of computational statistics, econometrics and machine learning: stochastic volatility, copulas, mixed-frequency (MIDAS) regression, jump-diffusion processes, and, since 2026, retrieval-augmented language models applied to corporate disclosure. Models are assessed out-of-sample wherever the data allow it.
Four active lines. Each pairs a modelling contribution with an out-of-sample test on market data.
Forecasting conditional variance and the far tail of the return distribution across calm and turbulent regimes, and turning those forecasts into Value-at-Risk and Expected Shortfall that hold up in backtesting. Realized EGARCH · empirical mode decomposition · periodic regime switching with jumps · Cornish–Fisher expansion
How the joint behaviour of spot and futures markets shifts with horizon and with market stress, and what that implies for the hedge ratio an investor should actually hold. Applied primarily to Taiwan index futures. Copula GJR-GARCH-MIDAS · conditional skewness and kurtosis · maximum-utility hedge ratios
Separating long-run from short-run volatility to isolate a climate-change risk premium, then testing whether that premium carries tradable information in a multi-asset global portfolio. GARCH-MIDAS · Realized Beta GARCH · variance and tail risk premia
Quantifying semantic turning points in annual reports with retrieval-augmented generation, and testing whether narrative shifts predict later operating performance and abnormal returns. Run jointly with colleagues in information engineering and design. RAG pipelines · LLM-based text quantification · event-study evaluation
A PI, four graduate researchers, and collaborators in information engineering and design.
Principal investigator
Associate Professor, Department of Finance, Da-Yeh University · PhD, Feng Chia University
Volatility modelling, dependence structure and futures hedging. Principal investigator on 17 competitive grants from Taiwan's National Science and Technology Council since 2009, three currently running. Published in the Journal of Banking & Finance, the Journal of International Money and Finance, the Journal of Empirical Finance and Mathematics and Computers in Simulation.
yhlai@mail.dyu.edu.tw · Office J302-01 · ext. 3524
Collaborator — Information Engineering
Assistant Professor, Department of Information Engineering, Da-Yeh University · PhD, National Chiao Tung University
Artificial intelligence and computational thinking. Co-author with the PI on studies of extreme return–volume dependence in the Taiwan market (2014) and of climate-change risk in long-run volatility and asset risk premia (2023), both conference papers. Collaborates on the machine-learning side of the lab's work on corporate disclosure.
Collaborator — Design
Assistant Professor, AI Innovation Design Program, Da-Yeh University · EdD, University of Exeter
Curriculum design, educational technology and applications of generative AI. Collaborates with the PI on cross-departmental teaching and on the design of the lab's public-facing work.
Cross-departmental teaching
AI Innovation Design Program, Da-Yeh University · 2026–27, semester 1
The three of us co-teach this course, which takes students from AI fundamentals and responsible use through to a capstone marketing project. It is where the lab's finance, information engineering and design strands meet in practice.
Graduate researchers
Department of Finance, Da-Yeh University · 2026–27
Doctoral and master's students supervised within the lab's areas of work. Graduate students have been co-authors on a number of the conference and journal papers listed below.
Competitive grants held as principal investigator. NSTC is Taiwan's national research funding agency.
| Period | Project | Funder | Status |
|---|---|---|---|
| 2026.08 – 2027.07 | Optimal hedge ratios under cross-excited jumps: a Hawkes jump-diffusion model and event-driven hedging strategies | NSTC | Active |
| 2026.01 – 2026.12 | Quantifying semantic shifts in corporate annual reports with generative AI and RAG: predicting operating performance and abnormal returns | MOE · TEEP | Active |
| 2025.08 – 2026.07 | Crash, fragility and co-moment risk in hedge ratios and hedging performance during market turbulence | NSTC | Active |
| 2023.08 – 2024.07 | Crash and fragility risk in Taiwan index futures hedging: a Copula Realized GARCH model with conditional skewness and kurtosis | NSTC | Complete |
| 2022.08 – 2023.07 | Climate change, risk premia and futures hedging effectiveness: a Double-Copula GJR-GARCH-MIDAS-M model | NSTC | Complete |
| 2021.08 – 2022.07 | Time-varying risk premia, risk aversion and maximum-utility hedge ratios: a Realized Beta GARCH model with variance and tail risk premia | NSTC | Complete |
| 2020.08 – 2021.07 | Adaptive expectations of long- and short-run volatility, hedging horizon and performance: a bivariate copula model with endogenous regime switching and MIDAS | NSTC | Complete |
| 2019.08 – 2020.07 | Excessive speculation, return volatility and optimal futures hedge ratios: a Realized Beta GARCH model with dynamic decay and jump-bias correction | NSTC | Complete |
| 2018.08 – 2019.07 | Forecasting Taiwan stock index return volatility with jumps and trading-session information: a periodic regime-switching approach | NSTC | Complete |
| 2017.08 – 2018.07 | Crash risk in the Taiwan stock market: the role of market liquidity and liquidity commonality | NSTC | Complete |
Nine NSTC projects listed above; a further eight, 2009–2017, are omitted for length.
Selected peer-reviewed articles. Conference papers and the full list are on the university profile.
Department of Finance
Da-Yeh University
Yi-Hao Lai, PhD
yhlai@mail.dyu.edu.tw
Office J302-01, Department of Finance
Da-Yeh University, Changhua 51591, Taiwan
Master's students interested in volatility modelling, hedging or financial NLP are welcome to write.