regime-detection

Classify market microstructure states per symbol on every tick.

Updated Mar 2, 2026

Search Tags:#risk-management #engine #bayesian #regime-detection #market-microstructure #tick-data #per-symbol

Ready to run

npx skills add https://github.com/Leiisawesome/feelies --skill regime-detection-leiisawesome

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: regime-detection
Source: https://github.com/Leiisawesome/feelies/tree/main/.cursor/skills/regime-detection
Command: npx skills add https://github.com/Leiisawesome/feelies --skill regime-detection-leiisawesome

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The platform-level regime detection service classifies market microstructure state per symbol on every tick, producing a posterior probability vector used by risk, sizing, and alpha layers to make regime-aware decisions.

Core Features & Use Cases

Online Bayesian regime inference: updates a posterior distribution per symbol each tick, based on NBBO quotes.
Single-writer, read-only consumers: orchestrator writes posterior; risk, sizing, and alpha components read current_state.
Deterministic fallbacks and fail-safes: neutral 1.0x defaults when data is missing or invalid; exact semantics defined by contract.

Quick Start

Configure the platform to feed NBBO quotes into the regime engine and watch the RegimeState posteriors drive risk and sizing decisions.

Frequently Asked Questions about regime-detection

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How does Bayesian regime detection work with tick data?▼

Online Bayesian regime detection updates a posterior probability vector per symbol on every tick using NBBO quotes. It classifies market microstructure states in real time to drive regime-aware risk and sizing decisions.

What is real-time market microstructure state classification?▼

It assigns a posterior probability vector to each symbol on every tick, classifying market microstructure states. This enables risk, sizing, and alpha layers to make regime-aware decisions based on current market conditions.

Can I run per-symbol regime detection across multiple tick data feeds?▼

Yes, the regime engine processes per-symbol tick data independently. An orchestrator writes posteriors while risk, sizing, and alpha components read the current state, ensuring deterministic updates across multiple feeds.

How do I handle missing or invalid tick data during regime inference?▼

The regime engine applies deterministic fallbacks with neutral 1.0x defaults when tick data is missing or invalid. This ensures stable regime-aware decisions without interrupting the posterior inference pipeline.

What is the best way to classify market microstructure states for risk management?▼

Using an online Bayesian regime engine that produces a posterior vector per symbol on every tick is an effective approach. It provides deterministic, idempotent updates and satisfies the RegimeEngine protocol for risk integration.