Researchers have developed a new hybrid model combining GARCH and LSTM for volatility forecasting, addressing a gap in assessing prediction uncertainty. The model extends existing GARCH-LSTM approaches by incorporating an ARMA structure for the conditional mean and introduces a forward bootstrap algorithm for multi-step-ahead prediction intervals. To improve interval coverage, the method draws bootstrap innovations from empirical distributions to mimic financial series' fat-tail properties and uses conformal prediction principles to systematically extend interval lengths. Evaluations in simulations and on U.S. stock indices from 2020-2025 indicate that the ARMA-based hybrid achieves lower point-forecast errors, and combining both remedies with the mean-based hybrid results in intervals closest to the nominal level. AI
IMPACT Introduces novel methods for uncertainty quantification in financial forecasting models, potentially improving risk management.
RANK_REASON The item is an academic paper detailing a new statistical model and its evaluation. [lever_c_demoted from research: ic=1 ai=0.4]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
