Chan Theory Quantitative Trading System (缠论量化交易系统)

A complete Python implementation of Chan Theory (缠中说禅) technical analysis — from raw K-line data to buy/sell signals, position management, and backtesting.

What is Chan Theory?

Chan Theory is a recursive geometric framework for analyzing financial markets, developed from 108 blog posts by "缠中说禅" (Chán Zhōng Shuō Chán). It decomposes market movement into self-similar structures at multiple levels:

K-line → Containment → Fractal → Bi (Stroke) → Segment → ZhongShu (Pivot Zone) → Trend

Key features:

• Recursive self-similarity: Structure at each level mirrors the levels above and below
• Complete classification: Every market move can be categorized as consolidation, uptrend, or downtrend
• MACD-area divergence: Measures move "force" as area rather than point-to-point comparison
• Three buy/sell point types: Each with defined entry conditions and stop-loss rules

Quick Start

pip install -e .

from chan.analyzer import ChanAnalyzer
from chan.data.fetcher import fetch_a_share_daily

# Fetch data (requires akshare)
bars = fetch_a_share_daily("000001", start_date="20230101")

# Run analysis
analyzer = ChanAnalyzer(min_signal_score=0)
result = analyzer.analyze(bars)

print(result.summary())
for signal in result.signals:
    print(signal)

Run the Example

PYTHONPATH=. python examples/basic_usage.py

启动前端诊断系统

streamlit run web_app.py

功能：输入股票或 ETF 代码，自动完成缠论结构分析，展示关键点位（ZG/ZD、近20K高低点）、趋势判断与买卖建议，并绘制包含分型/笔/中枢/信号的结构图。

Run a Backtest

from chan.backtest.engine import BacktestEngine, BacktestConfig

config = BacktestConfig(
    initial_capital=1_000_000,
    commission_rate=0.0003,
    symbol="000001",
)
engine = BacktestEngine(config)
result = engine.run(bars)
print(result.summary())

Architecture

Module	Purpose
chan/kline.py	K-line containment processing
chan/fractal.py	Top/bottom fractal identification
chan/bi.py	Bi (stroke) construction
chan/segment.py	Segment construction via characteristic sequence
chan/zhongshu.py	ZhongShu (pivot zone) identification
chan/macd.py	MACD calculation and area measurement
chan/divergence.py	Trend and consolidation divergence detection
chan/signals.py	Three-type buy/sell signal generation
chan/position.py	Progressive position management (1B→2B→3B)
chan/risk.py	Four-layer risk control system
chan/analyzer.py	Pipeline orchestrator
chan/backtest/	Backtesting engine and performance metrics
chan/data/fetcher.py	A-share data via akshare, CSV loading

Signal Types

Signal	Chinese	Trigger Condition
1B (First Buy)	一买	Trend bottom divergence (≥2 ZhongShu, MACD area declining)
2B (Second Buy)	二买	Pullback after 1B doesn't break 1B low
3B (Third Buy)	三买	ZhongShu breakout retest stays above ZG
1S (First Sell)	一卖	Trend top divergence
2S (Second Sell)	二卖	Bounce after 1S doesn't exceed 1S high
3S (Third Sell)	三卖	ZhongShu breakdown retest stays below ZD

Testing

pytest tests/ -v

Configuration

All parameters are centralized in chan/config.py. Key defaults:

• MACD: (12, 26, 9)
• Min Bi bars: 5 (new-bi rule)
• Divergence threshold: area ratio < 0.7
• Max position: 80% of capital
• Max single trade loss: 2% of capital

License

MIT