QuanuX Statistics Node

The QuanuX Statistics Node (quanux_stats) is a purpose-built service for real-time market analysis and signal generation. It bridges the gap between raw market data and high-level strategy logic by providing pre-calculated metrics.

Capabilities

1. Online Statistics:

- Uses Welford's Algorithm to track Variance and Standard Deviation with $O(1)$ complexity/storage per tick. - Tracks Z-scores relative to a rolling window (default 100 ticks). - Calculates Pairwise Correlation matrices in real-time.

1. Data Persistence:

- Ingests MARKET. data from NATS (JSON format). - Writes to DuckDB (market_stats.duckdb) for offline research and backtesting. - Future*: Will support Parquet export for S3 archival.

1. Signal Publishing:

- Publishes derived metrics to STATS.<SYMBOL> on NATS. - Format: {"symbol": "ES", "volatility": 12.5, "z_score": 1.2, ...}

Architecture

• Language: C++20
• Core: StatsEngine class.
• Storage: statsmap (Red-Black Tree) for per-symbol state.
• Concurrency: Single-threaded NATS consumer (for now) protected by std::mutex for future expansion.

Development Guide

Adding a New Metric

1. Update InstrumentStats in include/stats_engine.h:

Add a new accumulator (e.g., double sumlogreturn).

1. Update update() in src/stats_engine.cpp:

Implement the online update formula.

1. Publish:

Add the new field to the json derived object in the NATS callback.

Running Locally

# Build
cd QuanuX-Statistics/cpp/build
make quanux_stats

# Run (Ensure NATS is up)
./quanux_stats

Dependencies

• DuckDB: Embedded DB for storage.
• NATS C Client: High-performance messaging.
• Eigen: Linear algebra (Matrix operations).
• nlohmann/json: Serialization.