TX (Ternium S.A.) - Quantitative Investment Scorecard

Date: December 3, 2025 | Framework: Elements of Quantitative Investing

🎯 FINAL VERDICT: ACCEPTABLE HYPOTHESIS (B+ Grade)

Confidence Score: 80% | Expected Return: 23.13% | Optimal Position: 5%

📊 Quantitative Metrics Dashboard

Risk-Adjusted Performance

Overall Performance Grade: B+ (superior Sharpe, but high volatility and poor IR)

Factor Model Decomposition

Expected Return = 23.13% (annual)

  Risk-Free Rate:           +4.50%
  Market Premium (β=1.29):  +7.74%  ← Systematic risk
  Value Factor (P/B 0.61):  +6.00%  ← Largest alpha source
  Quality Factor (ROE 4.9): -2.50%  ← Earnings drag
  Momentum Factor:          +3.00%  ← Recent strength
  Dividend (risk-adjusted): +4.39%  ← After 80% cut probability
  ─────────────────────────────────
  TOTAL:                    23.13%

Key Driver: Value factor (+6%) is the engine. If mean reversion fails, return drops to ~12%.

Return Distribution (3-Year Monte Carlo)

Key Probabilities:

• Profit (any positive return): 65.5%
• Reach Graham Number ($63): 31.6%
• 50%+ gain: 39.0%
• 50%+ loss: 9.0%

Interpretation: Better than coin flip odds (65.5%) with attractive right tail (39% chance of 50%+ gain), but non-trivial left tail (9% chance of 50%+ loss).

Risk Metrics

Risk Grade: C (high volatility acceptable for value bet, but tail risk is brutal)

Position Sizing Analysis

Why 5% Despite Kelly Suggesting 87%?

1. Model uncertainty (cyclical timing risk 22-37%)
2. Idiosyncratic volatility (26% company-specific)
3. Dividend cut risk (80% probability)
4. Earnings quality issues (IR 0.416 < 0.50)
5. Behavioral risk (can you handle -50% drawdown?)

Sizing Grade: A (prudent restraint overrides mathematical excess)

🔬 Detailed Quantitative Assessment

Factor Loadings (Normalized)

Net Factor Profile: Leveraged value bet (1.50 loading) with quality concerns (-0.80) and momentum tailwind (0.70). High beta (1.29) amplifies all moves.

Profile Match: Similar to "distressed value" or "cyclical value" archetypes. Historically generates 20-30% returns in recovery phases, -30-50% drawdowns in extensions.

Earnings Quality Analysis

Composite Quality Grade: C- (balance sheet is A+, but earnings quality is D)

Key Finding: Q3 -88% miss is only 1.1 standard deviations from mean—meaning extreme volatility is NORMAL for TX, not an aberration. This is inherent to the cyclical steel business.

Dividend Sustainability Model

Expected Dividend Yield: 4.39% (vs 7.17% current) Shortfall: -2.78% from assumed yield Probability of ANY Cut: 80%

Grade: F (for current 7.17% sustainability, but C+ for risk-adjusted 4.39% expectation)

Cyclical Recovery Probability

Mean Reversion Model:

• Current ROE: 4.86%
• Normalized ROE: 15.0%
• Reversion Gap: 10.14%
• Half-Life: 2 years
• Formula: P(recovery) = 1 - e^(-t/2)

Grade: B+ (63-78% probability is favorable odds, but 22-37% failure risk is meaningful)

✅ What Passed Quantitative Tests (4/5)

Factor 1: Expected Return vs Risk-Free ✅ PASS

• Test: E(R) > Rf + 3%?
• Result: 23.13% > 7.50% ✅
• Margin: +15.63%

Factor 2: Sharpe Ratio vs Market ✅ PASS

• Test: SR > 0.80 × SR_market?
• Result: 0.532 > 0.266 ✅
• Margin: +100% vs threshold

Factor 3: Value Factor Exposure ✅ PASS

• Test: P/B < 0.70?
• Result: 0.61 < 0.70 ✅
• Value Loading: 1.50 (extreme)

Factor 4: Earnings Quality ❌ FAIL

• Test: Payout < 80% AND |Z-score| < 2?
• Result: 106.8% > 80% OR 1.06 < 2 ❌
• Issue: Unsustainable dividend

Factor 5: Optimal Position Size ✅ PASS

• Test: Half-Kelly ≥ 3%?
• Result: 87.39% ≥ 3% ✅
• Conclusion: Justifies meaningful allocation

Confidence Score: 80% (4/5 passed)

⚠️ Critical Red Flags

1. Dividend Cut Near Certainty (80% Probability)

• Current: 7.17% yield at 106.8% payout
• Expected: 4.39% yield after probable cut
• Impact: -2.78% on total return
• Action: Adjust return expectations downward

2. Tail Risk Underestimated by Report

• Report Claims: 20-25% max drawdown
• Quantitative Reality: 50-70% drawdowns (9-15% probability)
• Impact: Psychological preparation required
• Action: Position size discipline (5% max)

3. Earnings Quality is Poor (IR 0.416)

• Threshold: IR > 0.50 for institutional quality
• Actual: IR = 0.416
• Impact: -2.50% quality factor penalty
• Action: Accept this as cyclical nature, not fixable

4. Graham Number is NOT Base Case

• Report: "Base case"
• Quantitative: 31.6% probability (70th percentile)
• True Median: +27% over 3 years (+8.3% annual)
• Action: Temper expectations to median, not Graham

5. No Recovery Scenario (22-37% Probability)

• Probability: 22.3% no recovery in 3 years
• Impact: -30-50% returns if thesis fails
• Action: Use $32 stop loss religiously

✅ Critical Strengths

1. Superior Sharpe Ratio (0.532 vs 0.333)

• 60% better than market
• Validates risk-adjusted attractiveness
• Justifies allocation despite high volatility

2. Extreme Value Factor (P/B 0.61, Loading 1.50)

• Bottom decile valuation
• +6% value premium contribution
• Historical mean reversion supports recovery

3. Fortress Balance Sheet (D/E 13.9%, CR 2.46)

• Prevents bankruptcy in downturn
• Provides genuine downside protection
• Explains survival through -88% earnings miss

4. Cyclical Recovery Likely (63-78% probability)

• Better than coin flip odds
• 2-3 years is typical cycle timing
• Forward P/E 5.13x prices in recovery

5. Asymmetric Payoff (39% chance of 50%+ gain)

• Right tail is attractive
• 31.6% probability of Graham Number
• 90th percentile is +147% (3yr)

📋 Quantitative Recommendations

1. Investment Decision: PROCEED ✅

• Rationale: Sharpe 0.532 and expected return 23.13% justify allocation
• Condition: Must accept 35% volatility and dividend cut risk
• Profile: Enterprising value investors with 3+ year horizon

2. Position Size: 5% MAXIMUM 🎯

• Kelly suggests: 87.39%
• Quantitative caps: 5.00%
• Rationale: Model uncertainty, idiosyncratic risk, dividend cut
• Action: DO NOT EXCEED 5% regardless of conviction

3. Entry Strategy: Wait for Pullback ⏳

• Current Price: $37.68
• Target Range: $34-36
• Improvement: Sharpe increases from 0.532 to ~0.600
• Alternative: Enter 2.5% now, 2.5% at $34-36

4. Stop Loss: $32 (NON-NEGOTIABLE) 🛑

• Level: -15% from current
• Rationale: Below 200-day MA, signals thesis failure
• Action: Exit FULL position if triggered, no averaging down

5. Exit Triggers (Beyond Stop Loss) ⚠️

• Dividend cut >50%
• Earnings miss >50% for 2 consecutive quarters
• Payout ratio >120% for 2 consecutive quarters
• ROE <8% for 4 consecutive quarters

6. Expected Outcomes (Probabilistic) 📊

• Base Case (50%): +27% over 3 years (+8.3% annual)
• Favorable (75%): +86% over 3 years (+23.2% annual)
• Unfavorable (25%): -13% over 3 years (-4.4% annual)
• Probability of Profit: 65.5%

🎓 Lessons from Elements of Quantitative Investing

Applied Frameworks

1. Chapter 2: Univariate Returns

   • Historical volatility: 35% (steel sector realistic)
   • Return distribution: Fat tails, negative skew

2. Chapter 3: Sharpe Ratio

   • SR = (E[R] - Rf) / σ = 0.532
   • Cantelli's inequality: P(Loss > Lσ) ≤ 1/(1+(L+SR)²)

3. Chapter 4: Factor Models

   • r_TX = α + β·f_market + β·f_value + β·f_quality + β·f_momentum + ε
   • Factor decomposition: Market +7.74%, Value +6%, Quality -2.5%

4. Chapter 5: Evaluating Risk

   • VaR (95%): -34.44%
   • CVaR (95%): -49.06%
   • Z-score analysis: Q3 miss is 1.1σ (normal for TX)

5. Chapter 9: Portfolio Management

   • Kelly Criterion: f* = (E[R]-Rf)/σ² = 174.78%
   • Half-Kelly = 87.39%
   • Practical cap: 5% due to model uncertainty

🔄 Comparison to Report Conclusions

Net Assessment: Directional agreement (BUY at 5%), but quantitative analysis is more pessimistic on dividend and tail risk, yet more optimistic on expected return (if you accept the volatility).

🎯 Final Score: B+ (Good Hypothesis)

Why B+, Not A?

• Earnings volatility (Q3 -88% miss, 78.9% std dev)
• Dividend sustainability (106.8% payout, 80% cut probability)
• Quality factor negative (ROE 4.9%, IR 0.416)
• Tail risk severe (9% probability of 50%+ loss)

Why B+, Not C?

• Superior Sharpe (0.532 vs 0.333)
• Strong value exposure (P/B 0.61, loading 1.50)
• Cyclical recovery probable (63-78%)
• Balance sheet fortress (D/E 13.9%)
• Asymmetric upside (39% chance of 50%+ gain)

Confidence Score: 80%

4 out of 5 quantitative factors passed

🧠 The Psychology of This Investment

Can you handle:

1. ❌ -40% to -50% drawdowns (expected in bad years)
2. ❌ Dividend cut 50-100% (80% probability)
3. ❌ 3+ year holding period (recovery takes time)
4. ❌ 1-in-4 chance of losing money any given year
5. ❌ 1-in-10 chance of >50% loss over 3 years

If you answered NO to any: Reduce position to 3% or skip entirely.

If you answered YES to all: Proceed with 5%, use $32 stop loss, prepare mentally for volatility.

Remember: Expected return of 23.13% and Sharpe of 0.532 are mathematical constructs. Your actual returns depend on:

1. When you enter (tactical timing matters)
2. Whether you hold through -40% drawdowns (most don't)
3. Whether the cyclical recovery happens (63-78% probability)
4. Whether you exit at stop loss (discipline required)

📁 Generated Files

1. This Scorecard: /home/pengacau/pasar-malam/output/tx-quantitative-scorecard.md
2. Full Analysis (18k words): /home/pengacau/pasar-malam/output/tx-quantitative-verdict-2025-12-03.md
3. Executive Summary (2.5k words): /home/pengacau/pasar-malam/output/tx-quantitative-summary-2025-12-03.md
4. Report Comparison: /home/pengacau/pasar-malam/output/tx-report-vs-quantitative-comparison.md
5. Metrics JSON: /home/pengacau/pasar-malam/output/tx_quantitative_metrics.json
6. Analysis Script: /home/pengacau/pasar-malam/scripts/tx_quantitative_analysis.py

Analysis Completed: December 3, 2025 Framework: Elements of Quantitative Investing (Chapters 2-5, 9) Analyst: Senior Quantitative Investment Analyst

Disclaimer: This analysis is for educational purposes. Past performance does not guarantee future results. The quantitative models are based on historical data and assumptions that may not hold in the future. Invest at your own risk.