Forecasting Global Volatility with Predictive Spillover Networks: A Neuro-Econometric Spatio-Temporal Transformer for Asynchronous Financial Markets

Xinlin Zhao; Haotian Qiao

Home/Archive/arXiv

arXiv · 2608.25369

Forecasting Global Volatility with Predictive Spillover Networks: A Neuro-Econometric Spatio-Temporal Transformer for Asynchronous Financial Markets

Xinlin Zhao·Haotian Qiao

View original source ↗Read source document ↗Export metadata ↓

Abstract

Forecasting global realized volatility requires a model that can learn from interconnected markets without treating zero-coded exchange closures as observed zero volatility. We develop PGA-Trans-HAR, a neuro-econometric architecture that combines a rolling ridge-VAR/GFEVD predictive-connectedness network, masked spatio-temporal attention, and a frozen HAR anchor. Missing observations used to estimate the rolling econometric prior are completed only within the trailing information set available at the forecast origin. An asymmetric source mask prevents closed markets from transmitting zero-coded closure signals. A learned, market-specific gate allocates weight between the econometric prior and data-driven spatial attention. A bounded inverse-softplus correction then refines the HAR forecast while preserving positivity. We evaluate eight international equity indices from 2006 to 2022 at 1-, 5-, and 22-union-calendar-day forecast leads, where the target is the one-day realized volatility observed at the corresponding future date. The design uses five-seed ensembles, select-and-refit estimation, structural ablations, HAC-adjusted Diebold--Mariano tests, and block-bootstrap Model Confidence Sets. PGA-Trans-HAR records the lowest cross-market average MAE at the 1-day forecast lead and the lowest average MSE and MAE at the 5- and 22-union-calendar-day forecast leads. Relative to HAR, both losses decline for all eight markets at the 1- and 5-day forecast leads and for seven markets at the 22-day forecast lead. The evidence shows that combining an origin-aligned econometric network with masked attention can improve multi-market volatility forecasts in asynchronous financial environments.

Explore related subjects

cs.CE

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Xinlin Zhao, Haotian Qiao. 2026-09-05. Forecasting Global Volatility with Predictive Spillover Networks: A Neuro-Econometric Spatio-Temporal Transformer for Asynchronous Financial Markets. https://arxiv.org/abs/2608.25369

Cite the original work for its findings. Save a collection to share your selection of sources.

Discover connections

cs.CE subject

Connections use source metadata and explicit phrase matches, not verified experimental comparisons.

KEEP EXPLORING

Related papers

arXiv2026

SabreAgent: Language Models at Design Time for Lost-Sales Inventory Control

SabreAgent uses a language model at design time to construct two components for lost-sales inventory control: a product-specific seasonal prior and a validation-selected capped base-stock policy family. During operation, statistical forecasting and inventory optimization use these frozen artifacts to determine orders, with zero language-model calls. We evaluate the approach on the $1{,}320$ instances of InventoryBench. Under the benchmark's cost assumptions, the operations-research core draws on a zero-lead-time optimality result and a projected-inventory rule for positive deterministic lead times. The latter computes replenishment shortfalls by propagating inventory using sales along simulated demand paths. The seasonal prior adds forecast variants alongside the original forecaster, and the selected policy family handles stochastic lead times with order destruction. SabreAgent scores $0.6311$, compared with $0.5380$ for the strongest published baseline, and ranks first in all six benchmark cells. Ablations attribute most of the gain to the OR core. In the paired analysis, the seasonal component adds $1.79\%$ across the three real-data cells, and the search component adds $2.3\%$ across the two stochastic-lead-time cells. These results demonstrate how model-generated priors and policy structure can improve an OR controller through design-time use.

cs.CE↗

arXiv2026

Hierarchical Multi-Task Learning with Liquidity-Aware Signals for Stock Forecasting

Stock price forecasting is a long-standing challenge in computational finance, driven by the inherent randomness of markets and complex temporal patterns. While recent deep-learning models have raised forecasting accuracy by jointly modeling inter-stock and temporal price dynamics, they conflate inter-stock relationships with intra-stock temporal dependencies and focus solely on the univariate objective of price movement. To address these limitations, we propose LiMT, a Hierarchical Multi-Task Learning framework that integrates liquidity-aware signals for stock price forecasting. LiMT employs a Market Regime Encoder (MRE) module that first extracts contemporaneous cross-stock dependencies, then models each stock's temporal dynamics, yielding a unified latent state. Building on this latent state, we introduce a Liquidity-Driven Learning (LDL) module, a mixture-of-experts architecture that features cross-task gating mechanisms to jointly predict price movement, volatility, and trading volume. We further design an Adaptive Portfolio Optimization (APO) mechanism that converts multi-task forecasts into executable portfolio weights under transaction-cost and liquidity constraints. Extensive experiments on the CSI300 and CSI500 benchmarks show that LiMT performs best among strong neural and tree-based baselines across the reported metrics. In realistic CSI300 backtests, APO improves annualized return from 3.99% to 10.01% and Sharpe ratio from 1.22 to 1.86 over equal weighting, showing that the multi-task forecasts translate into deployable portfolio gains.

cs.CE↗

arXiv2026

Scientific capabilities and deployment sustainability of small-scale LLMs in biological wastewater treatment

Large language models (LLMs) are emerging as scientific assistants, yet their computational demands and limited domain specialization constrain sustainable deployment in environmental engineering. Here, we investigate whether domain-specialized small-scale LLMs can combine scientific capability with sustainable deployment in biological wastewater treatment. We developed a benchmark evaluating three scientific capabilities of LLMs: retrospective cognition, comprehension fidelity, and prospective extrapolation. BioWater (8 billion parameters, fine-tuned on specialized domain knowledge) achieved higher comprehension-fidelity scores than participating human experts and performance comparable to a 397-billion-parameter general-purpose LLM in retrospective cognition and prospective extrapolation. Human-BioWater collaboration generated a scientific hypothesis that was subsequently supported by laboratory experiments, demonstrating its potential to contribute to prospective scientific research. We further evaluated the economic and environmental implications of LLM deployment across global wastewater treatment plants (WWTPs). Locally deployed small-scale LLMs became more sustainable than cloud-based large-scale LLMs as inference demand increased in intelligent WWTPs. These findings highlight domain-specialized small-scale LLMs as a promising pathway towards scientifically capable, computationally efficient, and sustainably deployable artificial intelligence for wastewater treatment.

cs.CE↗