Riding the Waves: Smarter Risk Forecasting for Turbulent Markets
![During the 2008 global financial crisis, optimal window sizes for risk assessment dynamically adjusted alongside forecasts of Value at Risk [latex]VaR[/latex] and Expected Shortfall [latex]ES[/latex], demonstrating a responsive relationship between assessment parameters and evolving market conditions.](https://arxiv.org/html/2603.01157v1/2603.01157v1/x10.png)
A new adaptive window selection method improves financial risk predictions by dynamically adjusting to changing market conditions.
Science
Networked Risk: Forecasting Volatility in Financial Systems
A new study reveals how accounting for connections between financial institutions dramatically improves the accuracy of volatility predictions.
Beyond the Hype: Deep Learning’s Edge in Financial Forecasting
![Performance comparisons reveal that across models, a 10% volatility rescaling of gross profit and loss ([latex] PnL [/latex]) consistently demonstrates discernible differences in performance metrics.](https://arxiv.org/html/2603.01820v1/2603.01820v1/Images/Gross_PnL.png)
A rigorous new benchmark reveals which deep learning architectures consistently deliver superior risk-adjusted returns in financial time series prediction.
Reading Between the Lines: AI Models Predict Suicide Risk by Simulating Online Dialogue
A new framework leverages multi-agent systems and causal reasoning to better identify individuals at risk of suicide based on their online conversations.
Seeing the Forest and the Trees: AI Predicts Patient Outcomes from Digital Pathology
A new approach leverages the power of graph neural networks to analyze whole-slide images and predict patient survival rates with improved accuracy.
Taming Lattice QCD with Machine Learning
![A hybrid gauge-fixing approach, leveraging trained parameters from the L21S2-1N-Z scheme, demonstrates comparable performance to a pure iterative baseline-achieving a normalized computational cost of 0.9753-while exhibiting consistent convergence, as evidenced by the evolution of the gauge-fixing functional [latex]F[g][/latex], relative differences [latex]\Delta F[g][/latex], and the diminishing count of incomplete configurations across numerous test configurations.](https://arxiv.org/html/2602.23731v1/2602.23731v1/x8.png)
A new approach leverages convolutional neural networks to significantly speed up the complex process of gauge fixing in lattice quantum chromodynamics.
Taming Turbulence: AI-Powered Data Assimilation for Shock Waves
![The study demonstrates rapid and sustained correction of a two-dimensional blast wave simulation through a neural Ensemble Kalman Filter, achieving agreement with a reference solution within [latex] 4.0 \times 10^{-3} [/latex] time units and maintaining accuracy across five successive data assimilation steps up to [latex] 2.0 \times 10^{-2} [/latex] time units, as evidenced by the convergence of the ensemble mean and farthest ensemble member towards the established solution.](https://arxiv.org/html/2602.23461v1/2602.23461v1/x13.png)
Researchers have developed a new method combining neural networks with ensemble Kalman filtering to significantly improve the accuracy and stability of simulations involving compressible flows with shocks.
Bringing Order to Chance: Normalizing Stochastic Automata
New research reveals a method for converting weighted automata into equivalent probabilistic models, bridging a gap between these formalisms and offering a unified approach to analysis.
Predicting Power Grid Failures Before They Happen
A new data-driven approach forecasts high-impedance arc faults in medium-voltage distribution systems, potentially preventing costly outages and improving grid reliability.
Turning Crisis into Opportunity: How AI Can Unlock Hidden Knowledge
As organizations grapple with increasingly complex and interconnected crises, generative artificial intelligence offers a surprising path to innovation by repurposing existing knowledge assets.