Research insights on regime duration
Where detection once ended, duration now begins. The research team publishes findings on regime persistence, transition dynamics, and the machine learning architectures that estimate how long market conditions endure.
Latest research notes
Recent publications, model updates, and market commentaries from the quantitative research team. Each note reflects the current state of the forecasting pipeline and the questions that remain open.
Duration estimates widen as macro uncertainty climbs
The April model refresh shows expanded confidence intervals across equity and fixed income regimes. Central bank divergence, tariff policy shifts, and energy price volatility have increased the structural uncertainty component in duration forecasts. The research team examines whether this widening is transient or signals a more persistent change in regime behavior.
New attention architecture improves regime boundary detection
A transformer-based model variant now outperforms the existing LSTM ensemble on regime boundary identification by a measurable margin. The improvement is most pronounced in high-volatility transitions where temporal dependencies are complex. The team is validating the architecture before promoting it to the production pipeline.
White papers and deep research
Long-form studies on methodology
The research program publishes white papers on regime persistence modeling, duration forecasting architectures, and the integration of macroeconomic context into machine learning pipelines. Recent studies examine survival analysis for financial time series, attention mechanisms for regime boundary detection, and Bayesian approaches to uncertainty quantification in duration estimates. Each paper includes full methodology, data descriptions, and replication materials for independent validation. The team welcomes collaboration requests from academic researchers and institutional quant teams.