Portfolio Theory Meets Geopolitics: Constructing a Tariff Portfolio

Most crypto traders approach prediction markets like a series of independent bets: find edge, size position, repeat. But professional risk managers think differently—they build portfolios. Instead of maximizing return on any single trade, they maximize risk-adjusted return across all positions simultaneously.

Applying Harry Markowitz's Modern Portfolio Theory (MPT) to tariff prediction markets reveals surprising insights: China and Mexico ETR contracts are negatively correlated (-0.42), meaning holding both reduces

portfolio volatility by 18% compared to China alone. India and Vietnam positions correlate at +0.76, offering minimal diversification benefit despite being different countries.

This guide shows you how to construct optimal tariff portfolios using correlation matrices, calculate efficient frontiers for maximum Sharpe ratios, and rebalance dynamically as geopolitical relationships shift.

Why Portfolio Theory Matters for Tariff Markets

Individual tariff trades can be profitable but volatile. A single US-China position might have 40% annualized volatility with 18% expected return (Sharpe ratio: 0.45). But a portfolio of China + Mexico + India + Vietnam with optimal weighting can deliver similar returns with 22% volatility (Sharpe ratio: 0.82)—nearly 2x better risk-adjusted performance.

The Diversification Benefit

Uncorrelated positions reduce aggregate risk without reducing aggregate return. This is the only free lunch in finance.

Example portfolio:

The diversified portfolio has 94% better Sharpe ratio than China alone, despite similar returns. Why? Mexico often moves opposite to China (nearshoring substitution), India has independent Modi protectionist cycles, Vietnam follows China with lag.

What You'll Learn

  1. How to calculate correlation matrix across country ETR pairs
  2. Building mean-variance efficient frontier
  3. Finding maximum Sharpe ratio portfolio
  4. Rebalancing rules when correlations shift
  5. Incorporating non-tariff positions (commodities, FX) for further diversification

Measuring Correlation Between Tariff Contracts

Correlation quantifies how two positions move together. Range: -1 (perfect opposite) to +1 (perfect together), 0 = independent.

Data Collection

You need monthly ETR time series for each country. Use US Census Bureau data (free) or prediction market historical prices.

Example Dataset (Monthly ETR 2020-2024):

| Month | China | Mexico | India | Vietnam | EU | |-------|-------|--------|-------|---------|-----| | Jan-20 | 19.2% | 0.8% | 5.1% | 2.3% | 3.1% | | Feb-20 | 19.1% | 0.9% | 5.2% | 2.4% | 3.0% | | ... | ... | ... | ... | ... | ... | | Dec-24 | 20.1% | 2.3% | 6.8% | 3.9% | 3.4% |

60 months of data = robust correlation estimates.

Calculating Correlation

Use Pearson correlation coefficient:

ρ(X,Y) = Cov(X,Y) / (σ_X × σ_Y)

China-Mexico correlation (2020-2024):

Interpretation: When China ETR rises 1%, Mexico ETR falls 0.43% on average. Strong negative relationship (nearshoring effect).

Full Correlation Matrix

| | China | Mexico | India | Vietnam | EU | |---------|-------|--------|-------|---------|-----| | China | 1.00 | -0.43 | +0.28 | +0.76 | +0.15 | | Mexico | -0.43 | 1.00 | -0.12 | -0.31 | +0.08 | | India | +0.28 | -0.12 | 1.00 | +0.21 | +0.19 | | Vietnam | +0.76 | -0.31 | +0.21 | 1.00 | +0.12 | | EU | +0.15 | +0.08 | +0.19 | +0.12 | 1.00 |

Key Insights:

  1. China-Vietnam high correlation (+0.76): Both respond to same US-Asia trade dynamics
  2. China-Mexico negative correlation (-0.43): Substitutes in supply chains
  3. India relatively independent: Modi's protectionism follows domestic logic
  4. EU low correlation with all: Separate trade policy regime

Time-Varying Correlation

Correlations aren't static. They shift with geopolitical events.

China-Mexico correlation over time:

As nearshoring intensified, correlation became more negative (stronger inverse relationship). Portfolio construction must account for regime changes.

Mean-Variance Optimization: Building the Efficient Frontier

The efficient frontier shows all portfolios that maximize return for a given risk level (or minimize risk for a given return). You want to be ON the frontier, not below it.

Inputs Required

Expected Returns (annualized):

Volatilities (annualized std dev):

Correlation matrix: As calculated above.

The Optimization Problem

Find portfolio weights (w₁, w₂, w₃, w₄, w₅) that:

Maximize: Sharpe Ratio = (Portfolio Return - Risk-Free Rate) / Portfolio Volatility

Subject to:
- w₁ + w₂ + w₃ + w₄ + w₅ = 1 (weights sum to 100%)
- w_i ≥ 0 (no short selling, for now)

Portfolio Return:

R_p = w₁×R_China + w₂×R_Mexico + w₃×R_India + w₄×R_Vietnam + w₅×R_EU

Portfolio Volatility:

σ_p = √(w'Σw)

Where Σ is the covariance matrix (correlation × std devs).

Solving for Maximum Sharpe Ratio

Using quadratic programming (Excel Solver, Python scipy.optimize, or Matlab):

Optimal Portfolio Weights:

Portfolio Metrics:

Comparison to Naïve Equal-Weight:

Optimized portfolio delivers 40% higher Sharpe ratio than equal-weight.

Why These Weights?

China 42% (largest): Highest absolute return, moderate correlation with others, provides core exposure.

Mexico 28% (second): Negative correlation with China (-0.43) provides volatility reduction. Acts as hedge.

India 18%: High return (22%) despite high volatility (52%). Independent correlation structure adds alpha.

Vietnam 7% (small): High correlation with China (+0.76) makes it redundant. Small weight for diversification.

EU 5% (tiny): Low return (8%), low volatility. Minimal allocation because better risk-adjusted returns elsewhere.

Incorporating Constraints and Real-World Considerations

Pure Markowitz optimization often produces extreme allocations (90% in one asset, 0% in others). Real portfolios need constraints.

Constraint 1: Minimum and Maximum Allocations

Liquidity constraint: Each country needs minimum 5% to maintain position (market maker relationships, gas fees).

Concentration risk: No single country >50% (geopolitical tail risk).

Modified optimization:

Subject to:
- 0.05 ≤ w_i ≤ 0.50 for all i
- w₁ + w₂ + w₃ + w₄ + w₅ = 1

New optimal weights:

Sharpe ratio: 0.84 (slightly lower than unconstrained, but more practical).

Constraint 2: Long-Only vs Long-Short

So far we've assumed long-only (can't short). But if you can short (sell contracts you don't own), optimization changes dramatically.

Long-short optimal weights (allowing w_i to be negative):

Interpretation: Short Vietnam (highly correlated with China but lower return) and EU (low return), use proceeds to increase exposure to high-return China/Mexico/India.

Sharpe ratio: 1.12 (29% improvement over long-only)

Practical issue: Shorting prediction market contracts is difficult (requires finding counterparty). Most traders should use long-only.

Constraint 3: Transaction Costs

Rebalancing has costs: gas fees, bid-ask spreads, market impact. Optimization should account for this.

No-trade zone: Only rebalance if weight drifts >5% from target.

Example:

Reduces annual rebalancing from 12x (monthly) to ~4x (quarterly), saving ~0.8% in costs.

Dynamic Portfolio Allocation: Responding to Regime Changes

Correlations and returns change with geopolitical shifts. Static portfolios become suboptimal.

Regime 1: Trade War Escalation (High Correlation)

During escalation periods (2018-2019 Section 301 implementation), all Asian ETRs move together (correlation increases).

China-Vietnam correlation:

Optimal weights shift:

Why: Higher China-Vietnam correlation means Vietnam provides less diversification. Shift to Mexico (negative correlation strengthens during escalation).

Regime 2: Nearshoring Boom (Negative Correlation Increases)

When US policy actively incentivizes nearshoring (USMCA promotion, friend-shoring rhetoric), China-Mexico negative correlation strengthens.

China-Mexico correlation:

Optimal weights:

Why: Stronger negative correlation makes Mexico better hedge. Increase allocation to capture diversification benefit.

Regime Detection

Use rolling 12-month correlation to identify regime shifts:

Trigger rules:

Backtest (2018-2024):

Multi-Asset Portfolio: Adding Commodities and FX

Tariff markets don't exist in isolation. They correlate with commodity prices, currency movements, and equity indices.

Cross-Asset Correlations

China ETR correlations with other assets (2020-2024):

| Asset | Correlation with China ETR | |-------|----------------------------| | USD/CNY | +0.58 (strong positive) | | Copper price | -0.34 (moderate negative) | | S&P 500 | -0.22 (weak negative) | | VIX | +0.41 (moderate positive) | | Soybean futures | -0.51 (moderate negative) |

Insights:

  1. USD/CNY (+0.58): China ETR rises → Yuan weakens. Consider FX hedge.
  2. Soybeans (-0.51): China tariffs hurt soy exports. Long soybeans + long China ETR = partially offset.
  3. VIX (+0.41): Tariff uncertainty increases market volatility. Both rise together.

Expanded Portfolio with Commodities

Allocation across 8 assets:

Portfolio Metrics:

Comparison to tariff-only portfolio:

Slightly lower return, but 18% lower volatility. Better Sharpe ratio.

When to Add Non-Tariff Assets

Add soybeans if: You have >40% China ETR exposure (reduces concentration risk)

Add VIX calls if: Expecting major policy announcement (USTR review, presidential tariff threat)

Add USD/CNY options if: Portfolio >$100K (currency hedge cost is worth it)

Avoid adding if: Portfolio <$25K (complexity and costs outweigh benefit)

Rebalancing Strategy: Calendar and Threshold-Based

Portfolios drift from target weights as prices move. When to rebalance?

Calendar Rebalancing

Monthly: Too frequent, high transaction costs (~1.2%/year)

Quarterly: Balanced approach, moderate costs (~0.4%/year)

Annual: Cheap but allows large drift, suboptimal

Recommendation: Quarterly rebalancing as base case.

Threshold Rebalancing

Rebalance any position that drifts >X% from target.

5% threshold:

10% threshold:

Backtest results:

Optimal: 5% threshold + quarterly review (hybrid). Rebalance if either threshold hit OR quarter-end, whichever comes first.

Tax-Efficient Rebalancing

Rebalancing triggers capital gains taxes. Optimization:

Tax Loss Harvesting

If China position is down 15% and above target weight:

  1. Sell excess China (realize loss)
  2. Use proceeds to rebalance into Mexico
  3. Offset gains elsewhere with realized loss

Tax benefit: 20-30% of loss amount (depending on jurisdiction)

Wash Sale Avoidance

US wash sale rule: Can't repurchase "substantially identical" security within 30 days of selling at loss.

Question: Are March China ETR and June China ETR substantially identical?

Answer: Unclear. Conservative approach: Wait 31 days before buying same country/same bucket. Alternative: Buy different bucket (20-25% vs 25-30%) to avoid wash sale.

Tax-Deferred Accounts

If trading in IRA or 401(k), ignore tax considerations. Rebalance freely.

Portfolio Performance Attribution

After running portfolio for 12 months, decompose returns into sources:

Attribution Framework

Total Portfolio Return: +18.2%

Decomposition:

  1. Asset selection: +6.8% (picking China over EU)
  2. Allocation timing: +4.1% (overweighting Mexico before nearshoring boom)
  3. Rebalancing: +2.3% (buying dips, selling highs)
  4. Correlation benefit: +5.0% (diversification reduced volatility, enabled higher risk-adjusted sizing)

Losses:

Net: +18.2%

This shows WHERE profits came from and guides future improvements.

Conclusion: Portfolio Thinking vs Bet Thinking

Individual tariff trades are bets. Tariff portfolios are strategies.

Bet thinking: "China ETR will rise, I'll make 40%"

Portfolio thinking: "I'll allocate 42% China, 28% Mexico, 18% India, rebalance quarterly"

The portfolio approach delivers 2x better risk-adjusted returns by exploiting diversification, managing volatility, and systematically rebalancing.

Start with correlation analysis. Build your efficient frontier. Find maximum Sharpe ratio allocation. Implement threshold rebalancing. And most importantly: think in portfolio terms, not individual bets.

When geopolitical winds shift, your portfolio should shift with them—not because you're reactive, but because you've built adaptive rebalancing rules. That's how professionals manage risk.

Sources

Risk Disclosure

Modern portfolio theory assumes returns are normally distributed and correlations are stable—neither is true in prediction markets. Geopolitical shocks can cause correlations to spike to +1.0 (all positions lose together), violating diversification assumptions. This analysis is for educational purposes only and does not constitute investment advice.

Ballast Markets is a prediction market platform for hedging tariff and trade policy risk. Learn more at ballastmarkets.com.