Whitepapers & Methodology

Abstract

We present a physics-inspired framework for surfacing recurring events in financial time series. Rather than forecasting, the system observes, measures, and ranks historically recurring phenomena using three paradigms: (i) Signal Processing (price as a noisy signal), (ii) Complex Systems (markets as non-linear, phase-changing systems), and (iii) Hidden-State Estimation (observed price as a noisy measurement of an underlying state).

Events are defined as crossings between two mathematically constructed time series (e.g., an observation and a transform; a metric and a threshold). We quantify only historical properties: the number of events N, their day coverage (events observed on at least half the days in the window), and their excursions (historical movement following an event, measured deterministically between one crossing and the next). Our engine is indicator-agnostic, asset-agnostic within aligned markets, and non-advisory by design.