langsensei Sector Equity Scoring Skill
A reusable methodology for building quantitative value-scoring formulas for any A-share sector. Each formula produces a 0-100 composite score per stock, combining valuation, profitability, growth, cash flow, risk, and industry quality indicators via percentile ranking.
When to Use
- Building a new sector scoring formula from scratch
- Reviewing or updating an existing formula
- Comparing scoring approaches across sectors
Core Concepts
18-Indicator System
Every sector formula targets 18 indicators organized into 6 categories:
| Category | Standard Indicators | Typical Weight |
|---|---|---|
| Valuation (3) | PE-TTM, PB, Dividend Yield | ~15-25% |
| Profitability (5) | ROE, ROA, ROIC, Gross Margin, Net Margin | ~25-35% |
| Growth (2) | Revenue Growth YoY, Net Profit Growth YoY | ~10-25% |
| Cash Flow (3) | Operating CF / Net Profit, FCF Yield, Operating CF / Revenue | ~8-15% |
| Risk (3) | Debt-to-Asset Ratio, Current Ratio, Interest-bearing Debt Ratio | ~8-12% |
| Industry Quality (2) | Gross Margin YoY Change, Quick Ratio | ~5-10% |
Sector-specific substitutions are allowed. Financial sectors (banking, insurance, securities) replace inapplicable indicators with industry-specific ones (e.g., NPL ratio, provision coverage, net interest margin for banks; combined ratio, investment yield for insurance). The total must still target 18 indicators.
Percentile Ranking Method
Each indicator is scored by ranking all stocks in the sector and converting to a 0-100 percentile:
score_i = percentile within sector (0-100)
For "lower is better" indicators (PE, D/A, etc.):
score = 100 - percentile
For "higher is better" indicators (ROE, dividend yield, etc.):
score = percentile
percentile = (rank - 1) / (N_valid - 1) * 100
Tie-breaking: When multiple stocks share the same indicator value, they receive the same rank (standard competition ranking). The next rank is offset by the number of tied stocks (e.g., if two stocks tie at rank 3, the next stock receives rank 5).
Edge case — N_valid = 1: When only one stock has a valid value for an indicator, the percentile formula produces 0/0. In this case, assign a score of 50 (neutral) for that indicator.
The composite score is a weighted sum:
Value_Score = Sum(weight_i * score_i) / Sum(weight_i_participated)
Missing indicators are excluded — the denominator adjusts to only include weights of indicators with valid data.
Scoring Rules
These rules prevent outliers from distorting rankings:
| Rule | Description |
|---|---|
| Growth ±200% cap | Revenue and net profit growth rates are capped at ±200% before ranking. Display the capped value (e.g., 200.0, not the raw 1500%) |
| ROE/ROIC ≤60% cap | ROE and ROIC values exceeding 60% are capped at 60% before ranking |
| PE<0 exclusion | Loss-making stocks (negative PE) are excluded from PE ranking entirely (they receive no PE score) |
| Coverage penalty | If a stock has fewer than 10 valid indicators (out of 18), apply penalty: score *= valid_count / 18 |
| ST/*ST exclusion | All ST and *ST stocks are excluded from the ranking universe |
| B-share exclusion | B-share stocks (codes starting with 9xxxxx) are excluded |
FCF Yield Calculation
FCF_Yield = FCFF_BACK / f20
FCFF_BACK= Free cash flow to firm (from datacenter-web API: RPT_F10_FINANCE_MAINFINADATA)f20= Total market capitalization (from push2 batch API)- Do NOT use
f116(circulating market cap) — use total market cap
Weight Design Principles
Weights must sum to 100% and reflect sector economics. Use these guidelines:
Base Template
| Category | Default | Range |
|---|---|---|
| Valuation | 20% | 15-35% |
| Profitability | 30% | 25-35% |
| Growth | 15% | 10-25% |
| Cash Flow | 10% | 8-15% |
| Risk | 10% | 8-12% |
| Industry Quality | 8% | 5-10% |
There is no single correct weight allocation — it should be driven by sector characteristics.
Sector Adaptation Rules
| Sector Type | Adjustment | Rationale |
|---|---|---|
| Consumer staples (food, beverage, household) | Raise Profitability to ~35%, emphasize gross margin | Brand moat drives pricing power |
| Heavy-asset / PPP sectors (utilities, construction, transportation) | Raise Cash Flow to ~13-15%, raise Risk to ~12% | Capital-intensive, cash flow quality critical |
| Cyclical sectors (steel, coal, nonferrous, chemicals) | Lower Growth to ~10-12%, raise Valuation to ~25% | Growth is volatile and mean-reverting |
| Growth sectors (electronics, computer, pharma) | Raise Growth to ~20-25%, lower Valuation to ~15% | Revenue trajectory is the primary value driver |
| Financial sectors (banking, insurance, securities) | Replace Cash Flow/Risk with industry-specific indicators; Valuation ~30-35% | Standard cash flow metrics don't apply |
Indicator-Level Weight Guidance
Within each category, distribute weight based on indicator importance:
- Primary indicator (the most important metric): 40-60% of category weight
- Secondary indicators: 20-30% each
- No single indicator should exceed 15% of total weight (except in financials where PB for banks can be 15%)
- Risk category: keep each indicator at 2-5%, distribute evenly across 3 indicators
Standard Output Format
Every sector formula document follows this structure:
Part 1: Data Audit
- API data sources used (batch API for quotes, datacenter-web for financials)
- Indicator availability table: indicator name, API source, field name, coverage rate, data quality
- List of unavailable indicators (weight = 0%)
Part 2: Weight Table
- One table per category showing: indicator, weight, direction (higher/lower is better), notes
- Weight summary table: category, total weight, core logic explanation
- Weights must sum to exactly 100% (101% with rounding is acceptable)
Part 3: Formula
- The scoring formula (copy the standard percentile formula)
- Scoring rules applied (growth cap, ROE/ROIC cap, PE exclusion, coverage penalty)
- Any sector-specific formula adjustments
Part 4: Ranking Table
- Full ranking table with columns: Rank, Stock Name, Code, Score, and key indicator values
- Include an indicator coverage column (X/18 format showing how many indicators had valid data)
- Data date must be stated
- Total stock count and indicator count
Part 5: Reasonableness Checks
Perform all 7 standard checks:
| # | Check | What to Verify |
|---|---|---|
| 1 | Loss-making stocks | All PE<0 stocks should cluster near the bottom of rankings |
| 2 | High-score logic | Top-ranked stocks should have strong fundamentals across multiple categories, not just one extreme metric |
| 3 | Low-score logic | Bottom-ranked stocks should show clear weaknesses (losses, high debt, declining revenue) |
| 4 | Industry leader ranking | Well-known sector leaders should appear in a reasonable range (top quartile, though not necessarily #1) |
| 5 | Score distribution | Mean ~49-51, median ~49-52, >70 scores <15% of total, healthy spread |
| 6 | Cap validation | Verify that capped values appear correctly in ranking table (200.0 for growth, 60.0 for ROE/ROIC) |
| 7 | Sub-industry coverage | Check that the formula doesn't systematically favor one sub-industry over others within the sector |
Present checks as a verification table: dimension, expected, actual, conclusion (pass/fail).
Include a Limitations section noting known weaknesses (e.g., YoY vs CAGR, missing indicators, cross-sectional only).
Data Sources
| API | Endpoint | Purpose |
|---|---|---|
| Batch quote API | push2.eastmoney.com/api/qt/clist/get |
Sector stock list + basic indicators (PE, PB, market cap, dividend yield) |
| Financial data API | datacenter-web.eastmoney.com/api/data/v1/get |
Detailed financials per stock (RPT_F10_FINANCE_MAINFINADATA) |
Refer to the eastmoney-data skill for API usage details, field mappings, and request patterns.
Reference Formulas
See reference/a-share/ for 32 completed A-share sector formulas (01-banking through 32-conglomerate). These serve as concrete examples of the methodology applied to different sector types.