Crisis-Resilient Portfolio Management via Graph-based Spatio-Temporal Learning

TL;DR

CRISP is a graph-based spatio-temporal learning framework that adaptively models asset correlations during crises, enabling resilient portfolio management with interpretable regime detection and significant performance improvements.

Contribution

It introduces a dynamic, attention-based graph learning approach that discovers crisis-relevant asset relationships without fixed assumptions, improving predictive accuracy and robustness.

Findings

CRISP filters out 92.5% of irrelevant connections, focusing on crisis-relevant dependencies.

Achieves a Sharpe ratio of 3.76, outperforming baselines by over 90%.

Demonstrates strong generalization to different crisis regimes with interpretable regime detection.

Abstract

Financial time series forecasting faces a fundamental challenge: predicting optimal asset allocations requires understanding regime-dependent correlation structures that transform during crisis periods. Existing graph-based spatio-temporal learning approaches rely on predetermined graph topologies--correlation thresholds, sector classifications--that fail to adapt when market dynamics shift across different crisis mechanisms: credit contagion, pandemic shocks, or inflation-driven selloffs. We present CRISP (Crisis-Resilient Investment through Spatio-temporal Patterns), a graph-based spatio-temporal learning framework that encodes spatial relationships via Graph Convolutional Networks and temporal dynamics via BiLSTM with self-attention, then learns sparse structures through multi-head Graph Attention Networks. Unlike fixed-topology methods, CRISP discovers which asset relationships matter through attention mechanisms, filtering 92.5% of connections as noise while preserving crisis-relevant dependencies for accurate regime-specific predictions. Trained on 2005--2021 data encompassing credit and pandemic crises, CRISP demonstrates robust generalization to 2022--2024 inflation-driven markets--a fundamentally different regime--by accurately forecasting regime-appropriate correlation structures. This enables adaptive portfolio allocation that maintains profitability during downturns, achieving Sharpe ratio 3.76: 707% improvement over equal-weight baselines and 94% improvement over static graph methods. Learned attention weights provide interpretable regime detection, with defensive cluster attention strengthening 49% during crises versus 31% market-wide--emergent behavior from learning to forecast rather than imposing assumptions.