Stock Quantitative Analysis PRO: Dual Trend Resonance Stock Selection System

The Challenge of Trend Identification in Quantitative Stock Selection

In quantitative investing, relying on a single technical indicator (like a moving average crossover) for stock selection often leads to frequent false signals and lagging trend identification. Market noise can trigger invalid trades, while genuine trend movements require multi-dimensional signal resonance for confirmation. A robust system must not only detect patterns but also filter low-liquidity assets and assess the strength and sustainability of trends.

Core Capabilities and Workflow of the System

This skill employs a three-layer architecture to progressively filter stocks, ensuring those selected have a high probability of trending price action:

1. Liquidity Pre-filtering: By default, it enforces a hard threshold requiring daily trading volume greater than 100 million, eliminating assets with poor liquidity from the start to avoid difficulties in capital entry/exit. This step uses the amount field for filtering.
2. ADX Trend Strength Scoring: A native algorithm calculates the ADX value without relying on TA-Lib. The system assigns scores based on the ADX threshold (default >20):
   - ADX >40: Strong trend, +20 points
   - ADX 30-40: Medium trend, +15 points
   - ADX 20-30: Weak trend, +10 points
3. Technical Pattern Resonance Detection: Combines GMMA (Guppy Multiple Moving Averages) and Ichimoku Cloud for dual confirmation:
   - GMMA Bullish Alignment: Short-term moving average group > long-term group, with divergence >2%, ensuring the trend is diverging rather than converging, +30 points.
   - Cloud Analysis (9-26-52 Parameters): Price above the cloud's upper band (+20 points), conversion line > base line (+15 points), TK golden cross on the day (+5 points), cloud thickness >3% (+5 points).

Finally, the system aggregates all condition scores (out of 100) and outputs a S/A/B/C four-level signal rating, along with dynamically calculated stop-loss levels. For example, S-level (90+ points) indicates a strong recommendation, while C-level (<60 points) means the stock is dismissed.

Key Implementation Details and Applicability

• Data Requirements: At least 60 trading days of forward-adjusted K-line data (including OHLCV+Amount fields). Supports Tongdaxin local data (automatically scans vipdoc directories) and CSV file scanning.
• Adjustable Parameters: Liquidity threshold (liquidity) and ADX threshold (adx_threshold) can be tuned via interface parameters to adapt to different market conditions.
• Automatic Filtering: By default, it excludes stocks on the STAR Market (starting with 688) and BSE (starting with 8) to mitigate volatility risks.
• Caveats: This system is a quantitative analysis tool, and its outputs are for reference only. Markets carry systemic risks, and no signal should be construed as investment advice. Ensure data sources are accurate and timely when using this skill.