RollingCorrelationShiftDetector

Summary

RollingCorrelationShiftDetector is RTTA's streaming implementation of: Causal adjacent-window correlation shift detector.

Update API

result = rtta.RollingCorrelationShiftDetector(window=20, threshold=0.25).update(real0, real1)

The update(...) call consumes one observation using real0, real1. advance(...) uses the same inputs when the caller wants to update state without materializing a Python return value.

Theory Of Operation

RollingCorrelationShiftDetector compares two adjacent rolling windows: a reference window and a recent window. The C++ state moves expired recent samples into the reference window, maintains sufficient statistics, and emits the sign of the statistic difference when it exceeds the configured threshold.

Recurrence

Let \(z_t = (real0_t, real1_t)\) denote the observation consumed by one update(...) call and let \(\theta\) denote constructor parameters such as window lengths, thresholds, and smoothing constants.

\[\rho^R_t=\operatorname{corr}(R^x_t,R^y_t), \qquad \rho^B_t=\operatorname{corr}(B^x_t,B^y_t)\]

\[q_t=\rho^R_t-\rho^B_t\]

\[r_t = \begin{cases} 1, & q_t > h \\ -1, & q_t < -h \\ 0, & \text{otherwise} \end{cases}\]

The return value is the current scalar indicator value.

Implementation Notes

The recurrence is implemented in src/rtta/indicator.cpp in class RollingCorrelationShiftDetector.

Reference

Background reference