71_market_regime

71 — Market Regime Detector (ADX + Vol)

Category: Advanced Analytics / Real-Time

Computes ADX (Average Directional Index) and rolling volatility from live K-lines to classify the market regime: TRENDING, RANGING, or HIGH-VOL BREAKOUT.

How It Works

1. Fetches historical daily K-lines via request_history_kline (handles pagination).
2. Computes ADX using Wilder smoothing (pure stdlib — no TA-Lib).
3. Computes rolling annualised realised volatility.
4. Subscribes to CUR_KLINE for real-time bar updates.
5. On each new bar: recomputes ADX + vol, classifies regime:
   • TRENDING — ADX > 25 and DI ratio > 1.3
   • RANGING — ADX < 20
   • BREAKOUT — vol > 2× median and ADX rising
6. Prints formatted regime label every 30 seconds.

Usage

python3 main.py                         # defaults to HK.00700, ADX period=14
python3 main.py --stock HK.00700 --period 14 --window 100

Key Concepts Demonstrated

• Wilder smoothing (standard ADX calculation)
• CurKlineHandlerBase push subscription
• Rolling volatility via statistics.pstdev
• State classification with hysteresis thresholds
• No numpy/pandas required

Configuration

Flag	Default	Description
--stock	HK.00700	Stock ticker
--period	14	ADX period
--window	100	Rolling window for vol baseline