Susceptible Reservoir Architectures for Regime-Conditional Volatility Forecasting

Can quantum physics help us anticipate financial market stress? My ongoing research project, Susceptible Reservoir Architectures, explores financial volatility forecasting and stress-regime detection using (Quantum) Reservoir Computing.

Instead of trying to force a neural network to memorize market noise, SUSA utilizes complex-valued and open-system $q$-qubit reservoirs to extract hyper-sensitive features from financial time series. The key novelty (and in some sense the real breakthrough) is how SUSA interprets those features. In MoE varioantion of the architecture, we build a model that dynamically adapts its forecasting depending on whether the market is in a calm, onset, recovery, or persistent-stress state.