Uncertainty-Gated Generative Modeling
This work provides a significant improvement in risk-sensitive financial time-series forecasting for financial institutions and analysts, particularly in volatile markets.
This paper addresses the problem of overconfident models in financial time-series forecasting, which is prone to regime shifts and shocks. The proposed Uncertainty-Gated Generative Modeling (UGGM) significantly improves risk-sensitive forecasting, achieving a 63.5% MSE reduction on NYISO (0.3508 to 0.1281) and improved robustness under shock intervals (mSE: 0.2739 to 0.1748).
Financial time-series forecasting is a high-stakes problem where regime shifts and shocks make point-accurate yet overconfident models dangerous. We propose Uncertainty-Gated Generative Modeling (UGGM), which treats uncertainty as an internal control signal that gates (i) representation via gated reparameterization, (ii) propagation via similarity and confidence routing, and (iii) generation via uncertainty-controlled predictive distributions, together with uncertainty-driven regularization and calibration to curb miscalibration. Instantiated on Weak Innovation AutoEncoder (WIAE-GPF), our UG-WIAE-GPF significantly improves risk-sensitive forecasting, delivering a 63.5\% MSE reduction on NYISO (0.3508 $\rightarrow$ 0.1281), with improved robustness under shock intervals (mSE: 0.2739 $\rightarrow$ 0.1748).