Python sibling of the R lossratio package: loss-ratio analysis and projection for long-term health insurance. Stage-adaptive projection (exposure-driven before maturity, chain ladder after), with maturity point and cohort regime detection. In development.
ChronosDefrag is an experimental quantitative ML framework for temporal defragmentation: learning latent market regimes, reconstructing pseudo-timelines, and performing ultra-low-latency historical context routing using pure PyTorch tensor algebra without external vector databases.
