73_correlation_tracker

73 — Multi-Asset Correlation Tracker

Category: Advanced Analytics / Real-Time

Subscribes to 10+ tickers in parallel, maintains rolling 60-bar return series, computes a live Pearson correlation matrix, and flags pairs whose correlation has spiked or collapsed.

How It Works

1. User provides a ticker list (defaults to 10 HK blue-chips).
2. Subscribes to CUR_KLINE push for all tickers simultaneously.
3. Collects close prices and computes log-returns in rolling 60-bar windows.
4. Computes Pearson correlation matrix across all pairs (pure math — no numpy).
5. Compares current correlations to a 20-bar baseline window.
6. Flags pairs where correlation has shifted by more than ±0.3.
7. Prints formatted matrix every 30 seconds + real-time spike alerts.

Usage

python3 main.py
python3 main.py --tickers HK.00700 HK.09988 HK.03690 US.TCEHY
python3 main.py --window 100 --min-bars 30

Key Concepts Demonstrated

• Multi-ticker batch subscription via subscribe(code_list=[...])
• CurKlineHandlerBase push handler
• Pure-Python Pearson correlation (math.sqrt, sum)
• Rolling window via collections.deque(maxlen=...)
• Spike detection via baseline comparison

Default Tickers

HK.00700  HK.09988  HK.03690  HK.02318  HK.00941
HK.01024  HK.02020  HK.09618  HK.03692  HK.01810