Skip to content

Investment Methodology

This document explains FinWiz's comprehensive investment methodology, philosophy, and decision-making framework.

Overview

FinWiz employs a quantitative, rules-based investment methodology that combines fundamental analysis, technical analysis, and risk assessment to generate objective, data-driven investment decisions. The system emphasizes transparency, repeatability, and adaptability while maintaining strict quality standards.

Core Philosophy: Use deterministic Python calculations for speed and cost efficiency, reserve AI for complex reasoning that requires contextual understanding.

Investment Philosophy

Fundamental Principles

  1. Data-Driven Decision Making
  2. Every recommendation backed by quantitative metrics
  3. No subjective judgment or gut feelings

  4. Transparent scoring methodology with full calculation lineage

  5. Quality Over Quantity

  6. Focus on exceptional companies with durable competitive advantages
  7. Adaptive weighting for quality companies (50/25/25 vs 40/30/30)

  8. Higher standards for BUY recommendations (score ≥0.85)

  9. Risk-Adjusted Returns

  10. Return potential evaluated relative to risk exposure
  11. Volatility, drawdown, and beta considered alongside fundamentals

  12. Sector-appropriate risk tolerance (e.g., tech vs utilities)

  13. Multi-Dimensional Analysis

  14. Fundamental (40%): Long-term business quality
  15. Technical (30%): Market sentiment and momentum

  16. Risk (30%): Downside protection and volatility

  17. Fail Fast Validation

  18. Reject invalid data at source (sanity checks)
  19. No garbage in, no garbage out
  20. Strict validation prevents misleading recommendations

Analytical Framework

Three-Pillar Approach

1. Fundamental Analysis (40% Weight)

Purpose: Evaluate long-term business quality and financial health.

Key Metrics:

  • ROE (Return on Equity): Capital efficiency and profitability
  • Excellent: ≥30% (1.0 score)
  • Acceptable: 10-30% (0.4-0.8 score)

  • Poor: <10% (0.2 score)

  • Debt-to-Equity Ratio: Financial leverage and risk

  • Excellent: ≤0.3 (1.0 score)
  • Acceptable: 0.3-2.0 (0.4-0.8 score)

  • Poor: >2.0 (0.2 score)

  • Revenue Growth: Business expansion trajectory

  • Excellent: ≥20% (1.0 score)
  • Acceptable: 0-20% (0.4-0.8 score)

  • Concerning: <0% (0.2 score)

  • Profit Margin: Pricing power and operational efficiency

  • Excellent: ≥20% (1.0 score)
  • Acceptable: 5-20% (0.4-0.8 score)
  • Poor: <5% (0.2 score)

Philosophy: Companies with strong fundamentals deserve premium valuations and weather downturns better.

2. Technical Analysis (30% Weight)

Purpose: Capture market sentiment, momentum, and timing signals.

Key Indicators:

  • RSI (Relative Strength Index): Overbought/oversold conditions
  • Optimal: 40-60 (neutral, sustainable)
  • Acceptable: 30-70 (moderate momentum)

  • Extreme: <30 or >70 (reversal risk)

  • Trend Analysis: Price relative to moving averages

  • Strong uptrend: Price >10% above 200-day MA
  • Sideways: Price within ±5% of 200-day MA

  • Downtrend: Price below 200-day MA

  • MACD (Moving Average Convergence Divergence): Momentum shifts

  • Positive crossover: Bullish signal
  • Negative crossover: Bearish signal

Philosophy: Even great companies can have poor entry/exit timing. Technicals help optimize timing.

3. Risk Assessment (30% Weight)

Purpose: Quantify downside risk and volatility exposure.

Key Measures:

  • Volatility (Annualized Std Dev): Price fluctuation magnitude
  • Low: ≤15% (defensive)
  • Moderate: 15-35% (balanced)

  • High: >35% (aggressive)

  • Maximum Drawdown: Worst peak-to-trough decline

  • Low: ≤10% (stable)
  • Moderate: 10-30% (typical)

  • High: >30% (volatile)

  • Beta: Market sensitivity

  • Defensive: <0.7 (less volatile than market)
  • Neutral: 0.7-1.2 (market-like)
  • Aggressive: >1.2 (more volatile than market)

Philosophy: Risk management is return management. Understand downside before chasing upside.

Quality Company Framework

Detection Criteria

A company qualifies as "quality" when meeting 2 of 3 criteria:

  1. High ROE (≥20%) - Superior capital efficiency
  2. Low Debt (Debt/Equity ≤0.5) - Financial stability
  3. Strong Margins (≥15%) - Competitive moat

Adaptive Weighting

Quality companies receive 50/25/25 weighting (fundamental/technical/risk) instead of standard 40/30/30:

Rationale:

  • Quality companies with durable advantages weather short-term volatility
  • Strong fundamentals more predictive of long-term returns
  • Short-term technicals and risk less relevant for quality franchises

Example: ASML (monopoly in EUV lithography) - High volatility (38.9%) acceptable given exceptional fundamentals (ROE 53.9%, Debt 0.14, Margins 29.4%)

Scoring Methodology

Composite Score Calculation

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

# Quality companies (2+ quality criteria met)
composite_score = (
    0.50 * fundamental_score +  # +10% emphasis on fundamentals
    0.25 * technical_score +    # -5% less weight on technicals
    0.25 * risk_score           # -5% less weight on short-term risk
)

Grading Scale

Grade Score Range Quality Level
A+ 0.95-1.00 Exceptional - Rare investment gems
A 0.85-0.95 Excellent - Strong buy candidates
B+ 0.80-0.85 Very Good - Solid opportunities
B 0.75-0.80 Good - Acceptable with minor concerns
C+ 0.70-0.75 Fair - Hold, monitor developments
C 0.65-0.70 Mediocre - Minimum acceptable
D 0.50-0.65 Poor - Sell, multiple red flags
F 0.00-0.50 Failing - Strong sell, critical issues

Recommendation Thresholds

  • BUY: Score ≥0.85 (A or better)
  • HOLD: Score 0.65-0.85 (B+ to C)
  • SELL: Score <0.65 (D or F)

Data Quality Standards

Validation Hierarchy

  1. Sanity Checks - Reject obviously invalid data:
  2. ROE exactly 0.0 or >200%
  3. Price ≤$0
  4. Debt/Equity >100

  5. Revenue growth outside -95% to +1000%

  6. Freshness Requirements:

  7. Financial data: <90 days stale
  8. Price data: <1 day stale

  9. Fundamental metrics: <1 quarter stale

  10. Source Priority:

  11. Primary: Calculated from actual financials
  12. Secondary: Vendor-provided (validated)
  13. Never: Placeholder or estimated data

Fail Fast Principle

Philosophy: It's better to reject analysis than provide misleading recommendations.

Implementation:

Python
# Example: Revenue growth validation
if not (−0.95 <= revenue_growth <= 10.0):
    raise CriticalFieldError("Revenue growth outside valid range")

if revenue_growth == stale_yfinance_value:
    # Calculate from actual financials instead
    revenues = financials.loc['Total Revenue']
    revenue_growth = (latest - previous) / previous

Result: ASML revenue growth corrected from 0.7% (stale) to 2.56% (calculated), changing score and grade.

Investment Decision Process

For New Positions

Text Only
1. Screen universe → Identify candidates with score ≥0.75
2. Deep analysis → Calculate composite score across 3 dimensions
3. Quality check → Detect if adaptive weights apply
4. Generate grade → Map score to letter grade
5. Recommendation → BUY (≥0.85), HOLD (0.65-0.85), SELL (<0.65)
6. Position sizing → Scale allocation based on conviction:
   - 0.95+: Full position (maximum conviction)
   - 0.85-0.95: 75-100% position (high conviction)
   - 0.75-0.85: 50-75% position (moderate conviction)

For Existing Positions

Text Only
1. Re-score holding → Calculate current composite score
2. Compare threshold → Check against 0.65 KEEP/SELL threshold
3. Decision:
   - Score ≥0.65 → KEEP (C grade or better)
   - Score <0.65 → SELL (D or F grade)
4. Monitor → Re-evaluate quarterly or on material news

Portfolio Construction

Diversification Guidelines:

  • Maximum 15-20 positions for meaningful diversification
  • No single position >15% of portfolio (concentration risk)
  • Sector limits: Max 30% in any single sector
  • Quality mix: Aim for 50%+ in quality companies (adaptive weights)

Rebalancing Triggers:

  • Score crosses threshold ±0.05 (e.g., 0.70 → 0.75 may warrant HOLD → BUY upgrade)
  • Quarterly review minimum
  • Material news events (earnings surprises, management changes)

Sector Considerations

Sector-Specific Characteristics

Different sectors have different "normal" profiles:

Sector Typical ROE Typical Debt Typical Volatility Notes
Technology 20-30% Low-Moderate 30-40% High growth, rapid innovation
Healthcare 15-25% Moderate 25-35% Regulatory risk, long R&D cycles
Financials 10-15% High 25-35% Leverage-based model
Consumer Staples 15-25% Moderate 15-25% Defensive, stable demand
Energy 10-20% High 35-50% Commodity price sensitivity
Utilities 8-12% High 15-25% Regulated, stable cash flows
Industrials 12-18% Moderate-High 25-35% Cyclical, capital-intensive
Real Estate 5-10% Very High 20-30% Leverage-heavy, interest rate sensitive

Future Enhancement: Sector-specific scoring adjustments (not yet implemented).

AI Minimalism Philosophy

Use Python For

✅ Deterministic Calculations:

  • Composite score computation
  • Grade assignment
  • Metric validation
  • Data transformation

Benefits: 50-100x faster, $0 cost, 100% repeatable

Use AI For

✅ Complex Reasoning:

  • Qualitative analysis (management quality, competitive position)
  • News sentiment analysis
  • Industry research and context
  • Report narrative generation

Benefits: Captures nuance, handles unstructured data, provides context

Cost-Performance Trade-off

| Task | Python | AI | Speedup | Cost Savings | |------|--------|----|---------|--------------| | Scoring | 50-100ms | 5-10s | 50-100x | 100% | | Validation | <1ms | N/A | N/A | 100% | | Calculation | <1ms | 2-5s | 1000x+ | 100% | | Analysis | N/A | 10-30s | N/A | - |

Performance Optimization

Batch Processing

For portfolios with 10+ holdings:

  • Sequential: 5-11 hours (66 holdings)
  • Batch (5 concurrent): 20-40 minutes (66 holdings)
  • Speedup: 10-20x

Data Pre-fetching

  • Fetch all ticker data upfront (2-5 seconds)
  • Cache results with TTL (45 minutes default)
  • Parallel API calls where possible

Caching Strategy

Python
# Cache hierarchy
1. Memory cache (fastest, session-scoped)
2. File cache (fast, persistent across sessions)
3. Hybrid cache (best of both)

# TTL configuration
CACHE_TTL=2700  # 45 minutes (covers typical portfolio analysis)

Limitations and Future Enhancements

Current Limitations

  1. Quantitative Only: No qualitative factors (management, brand, ESG)
  2. Backward-Looking: Historical data, not predictive
  3. Market Agnostic: Doesn't adjust for market conditions or cycles
  4. Sector Neutral: No sector-specific adjustments (yet)
  5. US-Centric: Optimized for US equities, limited international

Planned Enhancements

Near-Term (3-6 months):

  • Sector-specific scoring adjustments
  • ESG integration (environmental, social, governance)
  • Sentiment analysis from news/social media
  • Enhanced alternatives matching algorithm

Medium-Term (6-12 months):

  • Machine learning for threshold optimization
  • Adaptive sector weighting based on cycle
  • Options-implied volatility analysis
  • Peer comparison benchmarking

Long-Term (12+ months):

  • Macroeconomic factor integration
  • Alternative data sources (satellite, web traffic)
  • Real-time portfolio monitoring
  • Predictive modeling for forward-looking scores

Best Practices

For Analysts

  1. Validate Data: Always check data freshness and completeness
  2. Context Matters: Consider sector norms and market conditions
  3. Multiple Sources: Verify critical metrics with independent sources
  4. Document Assumptions: Track calculation inputs and logic
  5. Monitor Changes: Re-score on material events, minimum quarterly

For Portfolio Managers

  1. Systematic Process: Follow decision framework consistently
  2. Position Sizing: Scale by conviction level (score within BUY range)
  3. Diversification: Maintain sector limits and position size caps
  4. Rebalancing Discipline: Act on threshold crossings
  5. Risk Management: Respect SELL signals, don't average down on fundamentals

For Individual Investors

  1. Start Small: Begin with paper trading or small positions
  2. Understand Scores: Read methodology, don't blindly follow
  3. Risk Tolerance: Adjust thresholds for personal risk profile
  4. Time Horizon: Align strategy with investment timeframe
  5. Professional Advice: Consult qualified advisor for personalized guidance

Ethical Considerations

Transparency

  • All scoring logic is open and documented
  • No black-box algorithms or hidden factors
  • Full calculation lineage available for audit

Objectivity

  • Rules-based system eliminates human bias
  • Same methodology applied to all securities
  • No conflicts of interest (no positions in analyzed securities)

Limitations Disclosure

  • Clear about what system can and cannot do
  • Not financial advice, educational tool only
  • Users must make own investment decisions

Data Privacy

  • No personal data collected or stored
  • Analysis results private to user
  • No sharing of portfolio holdings

References

Academic Literature

  • Modern Portfolio Theory: Markowitz, H. (1952) - Portfolio Selection
  • Factor Investing: Fama, E. & French, K. (1992) - Multi-factor models
  • Quality Investing: Asness, C. et al. (2014) - Quality Minus Junk
  • Behavioral Finance: Kahneman, D. & Tversky, A. - Prospect Theory

Industry Practice

  • CFA Institute: Investment Analysis and Portfolio Management
  • GARP: Quantitative Investment Analysis
  • CAIA: Alternative Investments and Modern Portfolio Theory

Technical Resources

  • Python for Finance: McKinney, W. - pandas for financial analysis
  • Quantitative Trading: Chan, E. - Algorithmic trading strategies
  • Machine Learning: Hastie, T. et al. - Statistical learning methods

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