Deep Analysis Scoring System¶

This document explains the comprehensive scoring system used by FinWiz's Deep Analysis crew to evaluate investment opportunities.

Overview¶

The Deep Analysis scoring system is a deterministic, Python-based engine that analyzes assets across three dimensions:

Fundamental Analysis (40% weight) - Financial health and performance
Technical Analysis (30% weight) - Price trends and momentum
Risk Assessment (30% weight) - Volatility and downside risk

For quality companies with exceptional fundamentals, adaptive weights (50%/25%/25%) are automatically applied to emphasize stability over short-term volatility.

Scoring Architecture¶

Composite Score Calculation¶

Python

# Standard weighting (most companies)
composite_score = (
    0.40 * fundamental_score +
    0.30 * technical_score +
    0.30 * risk_score
)

# Adaptive weighting (quality companies)
composite_score = (
    0.50 * fundamental_score +  # +10% emphasis
    0.25 * technical_score +    # -5% reduction
    0.25 * risk_score           # -5% reduction
)

Quality Company Detection¶

A company qualifies as "quality" when it meets at least 2 of 3 criteria:

Criterion	Threshold	Rationale
High ROE	≥ 20%	Superior capital efficiency
Low Debt	Debt/Equity ≤ 0.5	Financial stability
Strong Margins	Profit Margin ≥ 15%	Competitive moat and pricing power

Examples:

ASML (ROE 53.9%, Debt 0.14, Margin 29.4%) → 3/3 criteria = Quality Company ✓
Apple (High ROE, High Margin, but Debt 1.52) → ⅔ criteria = Quality Company ✓
GES (ROE 7.56%, Debt 2.97, Margin 1.01%) → 0/3 criteria = Standard Weighting

Fundamental Analysis (40% Weight)¶

Evaluates long-term financial health and business quality.

Fundamental Metrics and Scoring¶

Metric	Excellent	Very Good	Good	Acceptable	Poor	Score Formula
ROE	≥30%	20-30%	15-20%	10-15%	<10%	1.0 / 0.8 / 0.6 / 0.4 / 0.2
Debt/Equity	≤0.3	0.3-0.5	0.5-1.0	1.0-2.0	>2.0	1.0 / 0.8 / 0.6 / 0.4 / 0.2
Revenue Growth	≥20%	12-20%	5-12%	0-5%	<0%	1.0 / 0.8 / 0.6 / 0.4 / 0.2
Profit Margin	≥20%	15-20%	10-15%	5-10%	<5%	1.0 / 0.8 / 0.6 / 0.4 / 0.2

Fundamental Score Calculation¶

Python

fundamental_score = (
    0.25 * roe_score +
    0.25 * debt_score +
    0.25 * growth_score +
    0.25 * margin_score
)

Key Principles¶

Tolerant of Mature Companies: 0-5% growth scores 0.4 (acceptable) instead of penalizing
ROE Sanity Check: Rejects ROE exactly 0.0 or >200% as data errors
Debt Context: Considers industry norms (tech vs industrials)
Revenue Growth Source: Calculated from actual financials, not yfinance's stale field

Technical Analysis (30% Weight)¶

Evaluates price momentum, trends, and market sentiment.

Technical Metrics and Scoring¶

Metric	Excellent	Very Good	Good	Acceptable	Poor
RSI (14)	40-60	30-40, 60-70	20-30, 70-80	10-20, 80-90	<10, >90
Trend	Strong uptrend	Moderate uptrend	Sideways	Moderate downtrend	Strong downtrend
MACD	Positive crossover	Positive	Neutral	Negative	Negative crossover

Technical Score Calculation¶

Python

technical_score = (
    0.40 * rsi_score +
    0.30 * trend_score +
    0.30 * momentum_score
)

Trend Classification¶

Strong Uptrend: Price >10% above 200-day MA (score: 1.0)
Moderate Uptrend: Price 5-10% above 200-day MA (score: 0.8)
Sideways: Price within ±5% of 200-day MA (score: 0.5)
Downtrend: Price below 200-day MA (score: 0.2-0.4)

Risk Assessment (30% Weight)¶

Evaluates volatility, downside risk, and market sensitivity.

Risk Metrics and Scoring¶

Metric	Very Low Risk	Low Risk	Moderate Risk	High Risk	Very High Risk
Volatility	≤15%	15-25%	25-35%	35-50%	>50%
Max Drawdown	≤10%	10-20%	20-30%	30-40%	>40%
Beta	0.7-1.0	0.5-0.7, 1.0-1.2	0.3-0.5, 1.2-1.5	0-0.3, >1.5	Negative

Risk Score Calculation¶

Python

risk_score = (
    0.35 * volatility_score +
    0.35 * drawdown_score +
    0.30 * beta_score
)

Key Adjustments¶

Cyclical Sectors: Volatility thresholds raised 5% (semiconductors, industrials naturally more volatile)
Market Correlation: Beta 0.8-1.2 considered normal (not penalized)
Downside Focus: Max drawdown weighted equally with volatility

Grading Scale¶

Grade	Score Range	Recommendation	Description
A+	0.95 - 1.00	BUY	Exceptional opportunity - all metrics excellent
A	0.85 - 0.95	BUY	Strong buy - excellent fundamentals and technicals
B+	0.80 - 0.85	BUY	Good buy - solid across all dimensions
B	0.75 - 0.80	BUY	Attractive - minor weaknesses acceptable
C+	0.70 - 0.75	HOLD	Hold - mixed signals, monitor closely
C	0.65 - 0.70	HOLD	Hold - significant concerns in 1-2 areas
D	0.50 - 0.65	SELL	Sell - multiple red flags
F	0.00 - 0.50	SELL	Strong sell - fundamental problems

Recommendation Thresholds¶

Python

if composite_score >= 0.85:
    recommendation = "BUY"
elif composite_score >= 0.65:
    recommendation = "HOLD"
else:
    recommendation = "SELL"

Portfolio Decision Thresholds¶

Python

# Portfolio holdings processor
if composite_score >= 0.65:
    decision = "KEEP"  # C grade or better
else:
    decision = "SELL"  # D or F grade

Data Quality and Validation¶

Critical Field Sanity Checks¶

All metrics undergo validation before scoring:

Field	Valid Range	Rejection Criteria
ROE	-50% to +200%	Exactly 0.0 or outside range
Current Price	>$0	≤ $0
Debt/Equity	0 to 100	< 0 or > 100
Revenue Growth	-95% to +1000%	Outside range or obviously stale
Volatility	0% to 500%	< 0 or > 500%
Beta	-5.0 to +10.0	Outside range

Revenue Growth Calculation¶

Problem: yfinance's revenueGrowth field is often stale or incorrect.

Solution: Calculate from actual financial statements:

Python

financials = ticker_data.financials
revenues = financials.loc['Total Revenue'].sort_index(ascending=False)
latest = revenues.iloc[0]
previous = revenues.iloc[1]
revenue_growth = (latest - previous) / previous

Example (ASML):

❌ yfinance field: 0.7% (incorrect)
✅ Calculated: (28.26B - 27.56B) / 27.56B = 2.56% (correct)

Error Handling¶

When data fails validation:

Strict Mode (VALIDATION_STRICTNESS=error): Halt analysis, raise exception
Warn Mode (VALIDATION_STRICTNESS=warn): Log warning, skip metric (default)
Off Mode (VALIDATION_STRICTNESS=off): Accept any data (development only)

Real-World Examples¶

Example 1: ASML (Semiconductor Equipment)¶

Metrics:

ROE: 53.9% (Excellent: 1.0)
Debt/Equity: 0.14 (Excellent: 1.0)
Revenue Growth: 2.56% (Acceptable: 0.4)
Profit Margin: 29.4% (Excellent: 1.0)
Volatility: 38.9% (Moderate: 0.6)

Quality Company Detection:

✓ High ROE (53.9% > 20%)
✓ Low Debt (0.14 < 0.5)
✓ Strong Margins (29.4% > 15%)
Result: 3/3 criteria → Adaptive weights applied

Scoring:

Text Only

Fundamental: 0.84 (A)
Technical: 0.72 (C+)
Risk: 0.58 (D)

# With adaptive weights (50/25/25):
Composite = 0.50 * 0.84 + 0.25 * 0.72 + 0.25 * 0.58 = 0.765

Grade: B
Recommendation: BUY

Insight: Despite high volatility (semiconductor sector), excellent fundamentals and quality company status justify BUY recommendation.

Example 2: GES (Guess - Retail Apparel)¶

Metrics:

ROE: 7.56% (Poor: 0.2)
Debt/Equity: 2.97 (Very Poor: 0.2)
Revenue Growth: 2.3% (Acceptable: 0.4)
Profit Margin: 1.01% (Very Poor: 0.2)
Altman Z-Score: 2.14 (Bankruptcy risk zone)

Quality Company Detection:

✗ Low ROE (7.56% < 20%)
✗ High Debt (2.97 > 0.5)
✗ Weak Margins (1.01% < 15%)
Result: 0/3 criteria → Standard weights

Scoring:

Text Only

Fundamental: 0.26 (F)
Technical: 0.72 (C+)
Risk: 0.58 (D)

# Standard weights (40/30/30):
Composite = 0.40 * 0.26 + 0.30 * 0.72 + 0.30 * 0.58 = 0.564

Grade: D
Recommendation: SELL

Insight: Poor fundamentals and bankruptcy risk override neutral technicals. Clear SELL signal.

Example 3: VOW.DE (Volkswagen - Automotive)¶

Metrics:

ROE: 3.6% (Very Poor: 0.2)
Debt/Equity: 1.30 (Poor: 0.4)
Revenue Growth: 2.3% (Acceptable: 0.4)
Profit Margin: 2.3% (Poor: 0.2)

Scoring:

Text Only

Fundamental: 0.26 (F)
Technical: 0.72 (C+)
Risk: 0.58 (D)

Composite: 0.494

Grade: F
Recommendation: SELL

Insight: Failed traditional automotive manufacturer with weak fundamentals.

Performance Characteristics¶

Speed¶

Python Scorer: 50-100ms per ticker
AI Scorer (deprecated): 5-10 seconds per ticker
Speedup: 50-100x faster

Cost¶

Python Scorer: $0 per analysis
AI Scorer (deprecated): $0.01-0.05 per analysis
Savings: 100% cost reduction

Accuracy¶

Deterministic: Same input always produces same output
Transparent: Every score has clear calculation lineage
Auditable: Full calculation metadata stored in lineage

Configuration¶

Environment Variables¶

Bash

# Validation strictness
VALIDATION_STRICTNESS=warn  # off, warn, error

# Enable deep analysis
DEEP_ANALYSIS_ENABLED=true

# Batch processing for portfolios
BATCH_PREFETCH_ENABLED=true
DEEP_ANALYSIS_BATCH_SIZE=5

Scoring Thresholds¶

Configurable in src/finwiz/scoring/scoring_thresholds.py:

Python

@dataclass
class ScoringThresholds:
    # Fundamental thresholds
    roe_excellent: float = 0.30
    roe_very_good: float = 0.20
    # ... more thresholds

    # Weighting (can be overridden for special cases)
    weight_fundamental: float = 0.40
    weight_technical: float = 0.30
    weight_risk: float = 0.30

Future Enhancements¶

Planned Features¶

Sector-specific scoring adjustments
ESG integration
Sentiment analysis from news/social media
Options-implied volatility analysis
Peer comparison benchmarking

Research Areas¶

Machine learning for threshold optimization
Adaptive sector weighting
Macroeconomic factor integration
Alternative data sources (satellite imagery, web traffic)

Recommendation Engine - BUY/HOLD/SELL logic
Investment Methodology - Overall philosophy
Data Quality - Validation rules
Risk Framework - Risk assessment details
Quantitative Analysis - Technical indicators

References¶

Academic Literature¶

Factor Investing: Fama-French five-factor model
Quality Metrics: Piotroski F-Score, Altman Z-Score
Risk Measures: VaR, CVaR, maximum drawdown
Technical Analysis: "Technical Analysis of the Financial Markets" by John Murphy

Industry Standards¶

CFA Institute: Global Investment Performance Standards (GIPS)
FINRA: Suitability and know-your-customer rules
SEC: Investment Advisers Act disclosures

Last Updated: November 2025 Version: 2.0 (Post-Adaptive Weights Implementation) Maintainer: FinWiz Development Team

Deep Analysis Scoring System¶

Overview¶

Scoring Architecture¶

Composite Score Calculation¶

Quality Company Detection¶

Fundamental Analysis (40% Weight)¶

Fundamental Metrics and Scoring¶

Fundamental Score Calculation¶

Key Principles¶

Technical Analysis (30% Weight)¶

Technical Metrics and Scoring¶

Technical Score Calculation¶

Trend Classification¶

Risk Assessment (30% Weight)¶

Risk Metrics and Scoring¶

Risk Score Calculation¶

Key Adjustments¶

Grading Scale¶

Recommendation Thresholds¶

Portfolio Decision Thresholds¶

Data Quality and Validation¶

Critical Field Sanity Checks¶

Revenue Growth Calculation¶

Error Handling¶

Real-World Examples¶

Example 1: ASML (Semiconductor Equipment)¶

Example 2: GES (Guess - Retail Apparel)¶

Example 3: VOW.DE (Volkswagen - Automotive)¶

Performance Characteristics¶

Speed¶

Cost¶

Accuracy¶

Configuration¶

Environment Variables¶

Scoring Thresholds¶

Future Enhancements¶

Planned Features¶

Research Areas¶

Related Documentation¶

References¶

Academic Literature¶

Industry Standards¶