Quantitative Analysis of Ternium S.A. (TX) Investment Hypothesis

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

Executive Summary: Quantitative Verdict

VERDICT: ACCEPTABLE HYPOTHESIS (with significant caveats) Confidence Score: 80% (4/5 quantitative factors passed)

The quantitative analysis broadly supports the TX investment thesis but reveals critical blind spots around dividend sustainability and earnings quality that the original report underweights. The numbers tell a more nuanced story than the narrative suggests.

1. Statistical Validation: Return Distribution

Volatility Analysis (Reference: Chapter 2 - Univariate Returns)

The report's Beta of 1.29 significantly understates total risk for cyclical steel stocks:

Key Finding: Steel stocks exhibit 35% annual volatility—nearly double the market's 18%. The report's 20-25% max drawdown estimate is understated; quantitative analysis suggests 70-105% potential drawdown in tail scenarios.

Expected Return: CAPM Framework (Reference: Chapter 4.1 - Factor Models)

E(R) = Rf + β(Rm - Rf)
E(R) = 4.50% + 1.29 × 6.00% = 12.24%

Report Claim: 15-20% expected return CAPM Baseline: 12.24% (before factor premia)

2. Multi-Factor Expected Return Analysis

Factor Model Decomposition (Reference: Chapter 4.1 - Factor Models)

The linear factor model for TX returns:

r_TX = α + β_market·f_market + β_value·f_value + β_quality·f_quality + β_momentum·f_momentum + ε

Quantitative Assessment vs Report

Report Range: 15-20% expected return Quantitative Model: 23.13% expected return

VERDICT: Report's lower bound (15%) is conservative; quantitative analysis supports the high end (20%) and suggests even higher potential returns IF earnings recovery materializes. However, this relies heavily on the value factor premium.

Critical Caveat: Value Premium Decomposition

The 6% value premium contribution assumes:

1. Mean reversion to Graham Number ($63.16) over 3 years = 18.79% annualized
2. P/B normalization from 0.61x to 1.00x = 64% upside
3. Forward P/E realization: EPS recovery from $2.90 to $7.34 (153% growth)

Implied Discount Rate: The Forward P/E of 5.13x implies a whopping 23.06% discount rate—suggesting the market prices in significant execution risk on the earnings recovery.

3. Mean-Variance Optimization & Position Sizing

Sharpe Ratio Analysis (Reference: Chapter 3.3 - Sharpe Ratio)

Sharpe Ratio = (Expected Return - Risk-Free Rate) / Volatility

VERDICT: At 0.532 Sharpe, TX offers superior risk-adjusted returns vs the market (0.333), validating the investment thesis from a mean-variance perspective.

Information Ratio (Alpha per Unit of Tracking Error)

• Alpha: 10.89% (stock-specific return above CAPM)
• Tracking Error: 26.19% (idiosyncratic volatility)
• Information Ratio: 0.416 (below 0.50 benchmark)

Interpretation: TX generates positive alpha, but with high idiosyncratic risk. The IR of 0.416 is below institutional quality (>0.50), reflecting the earnings volatility problem.

Optimal Position Sizing: Kelly Criterion

The Kelly Criterion provides the mathematically optimal allocation:

Kelly Fraction = (Expected Return - Rf) / σ²

CRITICAL FINDING: The Kelly Criterion suggests TX could support an 87% allocation given the expected return profile. However, this assumes:

1. No model error (unrealistic for cyclical stocks)
2. No liquidity constraints
3. No behavioral/career risk

The report's 5% recommendation is actually PRUDENT given:

• High earnings volatility (Q3 -88% miss)
• Dividend cut probability >60%
• Cyclical timing uncertainty
• Idiosyncratic risk (26% volatility)

Quantitative Verdict: The 5% position sizing is appropriate despite the Kelly math suggesting higher. This is a case where qualitative judgment correctly tempers quantitative excess.

4. Earnings Quality Assessment

Statistical Significance of Q3 2025 Miss (Reference: Chapter 5 - Evaluating Risk)

Earnings Surprise Analysis:

• Q3 2025: -88% miss ($0.10 vs $0.84 estimate)
• Z-score: -1.06 standard deviations
• Statistical Significance: Within normal variance (<2σ)

Surprising Finding: While the -88% miss feels catastrophic, it's only 1.1σ from the mean given TX's historical earnings volatility. This stock exhibits extreme earnings variance as part of its cyclical nature.

Implication: The Q3 miss is not an outlier for TX—it's normal cyclical behavior. However, this also means earnings are unpredictable quarter-to-quarter.

Dividend Sustainability: Scenario Analysis

Current Payout Ratio: 106.8% (UNSUSTAINABLE)

CRITICAL DISAGREEMENT #1: The report assumes a 7.17% dividend yield in its 15-20% return calculation. The quantitative model assigns an 80% probability of dividend cut, reducing the expected yield to 4.39%—a 2.78% shortfall.

Risk-Adjusted Return Impact: Subtract 2.78% from report's expected return:

• Report Low (15%) → 12.22%
• Report High (20%) → 17.22%

Even after dividend haircut, returns remain attractive, but the income thesis is severely compromised.

Forward P/E Reality Check

Forward P/E = 5.13x implies Forward EPS = $7.34

Using Gordon Growth Model:

P = E × (1 + g) / (r - g)
$37.68 = $7.34 × 1.03 / (r - 0.03)
r = 23.06%

The market is pricing in a 23% required return—far above the CAPM 12.24%. This implies:

1. High probability of earnings NOT reaching $7.34
2. Significant execution risk on cyclical recovery
3. Structural concerns beyond cyclical timing

5. Risk-Adjusted Return Metrics

Value at Risk (VaR) Analysis (Reference: Chapter 3.3 - Cantelli's Inequality)

1-Year VaR at 95% Confidence:

• Parametric VaR: -34.44%
• Position size (5% of portfolio): Max loss = $1,722 on $5,000 position

3-Year Monte Carlo Simulation (10,000 trials):

• Median Return: +27.14%
• Mean Return: +41.05%
• 10th Percentile: -47.34% (bad outcome)
• 90th Percentile: +147.04% (good outcome)
• Probability of Loss: 34.49%
• Probability of 50%+ Gain: 39.01%

Cantelli's Inequality (distribution-free bound):

• P(Loss > 35%) ≤ 29.87%
• P(Loss > 52.5%) ≤ 19.49%
• P(Loss > 70%) ≤ 13.49%

Report Claim: "20-25% maximum drawdown" Quantitative Reality: Tail risk includes 50-70% drawdowns with ~15-20% probability over 3 years.

CRITICAL DISAGREEMENT #2: The report significantly understates tail risk. While the median outcome is attractive, the left tail is brutal for cyclical stocks.

Monte Carlo Outcomes Summary

6. Cyclical Regression & Recovery Probability

Mean Reversion Framework (Steel Sector)

Normalized ROE: 15.0% (historical average for steel sector) Current ROE: 4.86% Reversion Gap: 10.14%

Mean Reversion Half-Life: 2 years (typical for cyclical industries)

Recovery Probability:

• Within 1 year (2026): 39.3%
• Within 2 years (2026): 63.2%
• Within 3 years (2027): 77.7%

Report Claim: "2026-2027 cyclical recovery expected" Quantitative Assessment: 63-78% probability—PLAUSIBLE but NOT CERTAIN.

Implication: There's a 22-37% chance the recovery does NOT materialize in the report's timeframe. This is significant downside risk.

Cycle Timing Risk

Current cycle phase: TROUGH (since 2023) Time in downturn: ~2 years Typical cycle length: 3-5 years

Base Case: Recovery begins 2026 (3 years into downturn) ✓ Reasonable Bear Case: Structural decline, not just cycle (22% probability) Bull Case: Rapid recovery 2025-2026 (39% probability)

7. Monte Carlo Simulation: Key Insights

3-Year Holding Period Simulation

Distribution Characteristics:

• Fat Left Tail: Negative skewness from cyclical downturn scenarios
• Right Skew: Value unlocking if recovery happens
• High Kurtosis: Leptokurtic (fat tails both sides)

Outcome Probabilities:

• Reaching Graham Number: 31.6% (not the "base case"—it's a favorable scenario)
• Total loss >50%: 9.0% (non-trivial risk)
• Dividend cut: 80% (almost certain)

Quantitative Takeaway: The report presents the Graham Number recovery as "base case," but it's actually a 70th percentile outcome. The true base case (median) is +27% over 3 years, or ~8.3% annualized—still decent but far from the 15-20% claimed.

8. Key Disagreements with Original Report

Disagreement #1: Dividend Sustainability

Report Assumption: 7.17% yield maintained Quantitative Model: 4.39% expected yield (80% cut probability) Impact: -2.78% on expected return

Quantitative Verdict: The report materially overstates income from dividends. At 106.8% payout ratio, a cut is not just likely—it's inevitable without earnings recovery.

Disagreement #2: Tail Risk Understatement

Report Claim: "20-25% maximum drawdown risk" Quantitative Analysis: 50-70% drawdowns possible (15-20% probability)

Quantitative Verdict: The report presents a sanitized risk picture. Steel stocks are VOLATILE, and TX's 35% annual volatility implies brutal drawdowns in adverse scenarios.

Disagreement #3: Base Case Definition

Report: Reaching Graham Number ($63.16) is "base case" Quantitative: 31.6% probability = favorable scenario, not base case

Quantitative Verdict: The report's "base case" is actually a 60-70th percentile outcome. The median outcome (+27% over 3 years) is much more modest.

Disagreement #4: Expected Return Precision

Report: "15-20% expected return" Quantitative: 23.13% expected return (but with HUGE variance)

Quantitative Verdict: The mathematical expected return is actually HIGHER than the report claims, driven by the fat right tail (value unlocking scenarios). However, the variance is so high that the precision of "15-20%" is misleading. A more honest statement: "Expected return 20-25%, but with 35% of outcomes showing losses."

Agreement #1: 5% Position Sizing

Report: 5% for neutral investors Quantitative: 5% (Kelly suggests higher, but prudence dictates capping)

Quantitative Verdict: The report gets this RIGHT. Despite Kelly math suggesting 87% allocation, the model uncertainty, earnings volatility, and dividend risk justify conservative sizing.

Agreement #2: Entry Timing Caution

Report: "Wait for pullback to $34-36" Quantitative: Current $37.68 is near fair value, pullback would improve Sharpe

Quantitative Verdict: The report's tactical patience is justified. At $34-36, the margin of safety increases from 40% to 45-50%, improving risk/reward.

Agreement #3: Strong Balance Sheet

Report: "Fortress balance sheet" Quantitative: Debt/Equity 13.9%, Current Ratio 2.46 = EXCELLENT

Quantitative Verdict: This is accurately stated. The balance sheet provides genuine downside protection and explains why TX survives earnings collapses.

9. Factor Exposure Summary

Factor Loadings (Normalized)

Net Factor Exposure: TX is a leveraged value play with quality concerns. The bet is that value reversion overwhelms quality decay.

Historical Precedent: This factor profile (high value, low quality, high momentum) typically generates:

• 20-30% annualized returns in recovery phases (2-3 years)
• -30-50% drawdowns in downturn extensions (1-2 years)

Current Outlook: We're ~2 years into the downturn, suggesting favorable timing for entry if cycle recovery thesis holds.

10. Quantitative Recommendations

1. Investment Decision

PROCEED with investment, but with STRICT risk management

Rationale:

• Sharpe Ratio 0.532 > Market 0.333 ✓
• Expected return 23.13% is attractive ✓
• Value factor (P/B 0.61) offers structural edge ✓
• Confidence score 80% (4/5 factors passed) ✓

But with caveats:

• Dividend cut 80% probable (adjust expectations)
• Tail risk >50% loss is 9% (prepare psychologically)
• Earnings recovery is necessary (63-78% probable, not certain)

2. Position Sizing

MAXIMUM 5% of portfolio

Do NOT exceed 5% even if Kelly math suggests higher

Rationale:

• Idiosyncratic risk (26% volatility) too high for concentration
• Dividend uncertainty creates income volatility
• Model error for cyclical timing is substantial
• Career/behavioral risk: Can you stomach -50% drawdown?

3. Entry Strategy

Recommendation: Execute in TWO tranches

Tranche 1 (2.5%): Enter at current $37.68 if impatient Tranche 2 (2.5%): Wait for pullback to $34-36 (if it occurs)

Alternative: If patient, wait for $34-36 and enter full 5%

Stop Loss: $32.00 (strict adherence required)

• Represents -15% from current price
• Below 200-day MA
• Signals cycle timing thesis is wrong

4. Expected Outcomes (Probabilistic)

Blended Expected Return: 23.13% annualized over 3 years

5. Risk Management Rules (NON-NEGOTIABLE)

Automatic Exit Triggers:

1. Price falls below $32.00 (stop loss)
2. Dividend cut >50% announced
3. Two consecutive quarters with earnings miss >50%
4. Payout ratio exceeds 120% for 2 consecutive quarters
5. ROE remains below 8% for 4 consecutive quarters

Monitoring Requirements:

• Quarterly: Earnings reports, payout ratio, ROE trends
• Monthly: Steel price indicators, Latin America macro data
• Weekly: Price vs stop loss ($32.00)

6. Portfolio Context

Suitable For:

• Investors with 3+ year time horizon
• Those who can tolerate 35% annual volatility
• Value investors comfortable with cyclical timing risk
• Portfolios with <15% cyclical exposure already

NOT Suitable For:

• Income-focused investors needing stable dividends
• Conservative investors (failed defensive criteria)
• Those who can't tolerate >50% drawdowns
• Portfolios already heavy in materials/industrials

11. Comparison: Report vs Quantitative Model

Where They Agree ✓

Where They Disagree ⚠️

Net Assessment

Directional Agreement: Both report and quantitative analysis conclude TX is a BUY at 5% position size.

Magnitude Disagreement: Report is too optimistic on dividend income and too complacent on tail risk, while being too conservative on expected return (if you're willing to accept the volatility).

Risk Communication: The report's language ("substantial downside protection") is misleading. There IS downside protection from balance sheet, but NOT from price volatility. A better framing: "Balance sheet protects against bankruptcy, but expect brutal volatility."

12. Final Quantitative Verdict

THE NUMBERS

• Expected Return: 23.13% annualized (3-year horizon)
• Sharpe Ratio: 0.532 (60% better than market)
• Optimal Position Size: 5.00% (Kelly suggests higher, but prudence caps it)
• VaR (95%, 1-year): -34.44%
• Probability of Loss (1-year): 25.4%
• Probability of Profit (3-year): 65.5%
• Confidence Score: 80% (4/5 factors passed)

THE VERDICT

GOOD HYPOTHESIS — The TX investment thesis is quantitatively sound for a 5% allocation in a value-oriented portfolio with 3+ year horizon.

Why GOOD:

1. Superior risk-adjusted returns (Sharpe 0.532 vs Market 0.333)
2. Strong value factor exposure (P/B 0.61) offers reversion premium
3. Cyclical recovery probability 63-78% is favorable odds
4. Balance sheet provides genuine downside bankruptcy protection
5. 5% position sizing appropriately balances opportunity vs risk

Why NOT GREAT:

1. Dividend cut probability 80% destroys income thesis
2. Earnings quality is POOR (Information Ratio 0.416)
3. Tail risk (50-70% drawdowns) is 9-15% probable
4. Cyclical timing uncertainty (22-37% no recovery)
5. High idiosyncratic volatility (26%) limits concentration

THE PSYCHOLOGY

Can you handle:

• Watching your position drop 40-50% in a bad year?
• Dividend being cut 50-100% (80% probability)?
• Waiting 3+ years for recovery that might not come (22% risk)?

If YES: Proceed with 5% position If NO: Reduce to 3% or skip entirely

THE MATH SAYS GO, BUT...

The quantitative analysis supports the investment, but with a critical caveat: The report's narrative is too sunny. This is not a "margin of safety + income" story. It's a "high risk, high return cyclical value bet" story.

Reframe the thesis: You're betting $5,000 (5% of $100k portfolio) that:

1. Steel cycle recovers 2026-27 (63-78% probability)
2. EPS recovers from $2.90 to $5-7 range (required for forward P/E)
3. You can stomach 35% annual volatility and potential -50% drawdown
4. You're okay with dividend being cut (near certainty)

If those are acceptable: The expected 23% return and 0.532 Sharpe make this a compelling quantitative opportunity.

If those are unacceptable: The numbers don't matter—skip the position.

QUANTITATIVE VS QUALITATIVE

Hold the numbers stronger: The quantitative expected return (23.13%) and Sharpe (0.532) are superior to qualitative judgment. Graham would approve of the value (P/B 0.61), but would be troubled by the earnings instability and dividend unsustainability.

Where quant wins: Expected return calculation, risk-adjusted performance, optimal sizing Where qualitative wins: Dividend cut assessment (numbers support the skepticism), tail risk communication

13. References from Elements of Quantitative Investing

This analysis applies the following frameworks:

1. Chapter 2: Univariate Returns — Historical volatility estimation, return distribution modeling
2. Chapter 3: Expected Return & Performance — CAPM, expected return calculation
3. Chapter 3.3: Sharpe Ratio — Risk-adjusted return analysis, Cantelli's inequality for tail risk
4. Chapter 4.1: Factor Models — Multi-factor return decomposition (r_t = α + B·f_t + ε_t)
5. Chapter 5: Evaluating Risk — Covariance matrix estimation, VaR calculation
6. Chapter 9: Portfolio Management — Mean-variance optimization, Kelly Criterion position sizing

Key Formulas Applied

1. Factor Model:

r_TX = α + β_mkt·r_mkt + β_value·r_value + β_quality·r_quality + β_momentum·r_momentum + ε

2. Sharpe Ratio:

SR = (E[R] - Rf) / σ = (23.13% - 4.50%) / 35.0% = 0.532

3. Kelly Criterion:

f* = (E[R] - Rf) / σ² = (23.13% - 4.50%) / (0.35)² = 174.78%
Half-Kelly = 87.39%

4. VaR (95% confidence):

VaR = E[R] - z_0.95 · σ = 23.13% - 1.645 · 35.0% = -34.44%

5. Cantelli's Inequality:

P(R < -L·σ) ≤ 1 / (1 + (L + SR)²)

14. Conclusion: The Quantitative Truth

The TX investment hypothesis is mathematically sound but psychologically challenging.

The quantitative case FOR:

• Expected return 23.13% is excellent
• Sharpe 0.532 beats market by 60%
• Value factor offers reversion premium
• 65% probability of profit over 3 years
• Balance sheet prevents bankruptcy

The quantitative case AGAINST:

• 35% volatility = brutal drawdowns
• 80% dividend cut probability
• Earnings quality is poor (IR 0.416)
• 9% probability of >50% loss
• Cyclical timing risk (22-37% no recovery)

The synthesis: If you can stomach the volatility and accept the dividend will be cut, the risk-adjusted returns justify a 5% position. But this is NOT a "sleep well at night" investment—it's a calculated high-risk, high-return bet on cyclical recovery.

Quantitative grade: B+ (Good Hypothesis)

• Not an A because of earnings volatility and dividend sustainability
• Not a C because the Sharpe ratio and value factor are compelling
• The B+ reflects "good opportunity with meaningful risks"

Graham would say: "The margin of safety is there (40%), but the business quality is not. Only for enterprising investors willing to do the work and tolerate the volatility."

The quant says: "The numbers check out. Sharpe 0.532, expected return 23%, 80% confidence score. But position size discipline is CRITICAL—don't exceed 5%, and use that $32 stop loss religiously."

Appendix: Files Generated

1. Quantitative Analysis Script: /home/pengacau/pasar-malam/scripts/tx_quantitative_analysis.py
2. Quantitative Metrics JSON: /home/pengacau/pasar-malam/output/tx_quantitative_metrics.json
3. This Report: /home/pengacau/pasar-malam/output/tx-quantitative-verdict-2025-12-03.md
4. Original Report: /home/pengacau/pasar-malam/output/tx-2025-12-03.md
5. Market Data: /home/pengacau/pasar-malam/output/tx_data_summary.md

Analysis Date: December 3, 2025 Analyst: Senior Quantitative Investment Analyst Framework: Elements of Quantitative Investing (Chapters 2-5, 9) Disclaimer: This analysis is for educational purposes. Past performance does not guarantee future results. Invest at your own risk.