foxhunt/risk/tests/portfolio_optimization_tests.rs

//! Portfolio Optimization Comprehensive Test Suite
//!
//! Tests cover:
//! - Mean-variance optimization (Markowitz)
//! - Kelly criterion (full and fractional)
//! - Risk parity optimization
//! - Black-Litterman model
//! - Efficient frontier construction
//! - Edge cases: empty portfolios, singular matrices, highly correlated assets
//! - Constraint handling: long-only, position limits, leverage
//! - Transaction cost impact
//! - Numerical stability with ill-conditioned matrices

#![allow(unused_crate_dependencies)]

use approx::assert_relative_eq;
use risk::portfolio_optimization::{OptimizationMethod, PortfolioConstraints, PortfolioOptimizer};

// ==================== HELPER FUNCTIONS ====================

/// Create a simple 3-asset portfolio for testing
fn create_simple_portfolio() -> PortfolioOptimizer {
    let assets = vec!["AAPL".to_string(), "GOOGL".to_string(), "MSFT".to_string()];
    let returns = vec![0.10, 0.12, 0.08]; // 10%, 12%, 8% expected returns
    let covariance = vec![
        vec![0.04, 0.01, 0.02], // AAPL variance = 0.04 (20% vol)
        vec![0.01, 0.09, 0.01], // GOOGL variance = 0.09 (30% vol)
        vec![0.02, 0.01, 0.05], // MSFT variance = 0.05 (22% vol)
    ];

    PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02, // 2% risk-free rate
        PortfolioConstraints::default(),
    )
    .expect("Failed to create portfolio optimizer")
}

/// Create a 2-asset portfolio with no correlation
fn create_uncorrelated_portfolio() -> PortfolioOptimizer {
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.10, 0.15];
    let covariance = vec![
        vec![0.04, 0.00], // No correlation
        vec![0.00, 0.09],
    ];

    PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .expect("Failed to create uncorrelated portfolio")
}

/// Create a portfolio with highly correlated assets (multicollinearity)
fn create_correlated_portfolio() -> PortfolioOptimizer {
    let assets = vec!["X".to_string(), "Y".to_string(), "Z".to_string()];
    let returns = vec![0.10, 0.11, 0.12];
    let covariance = vec![
        vec![0.04, 0.038, 0.037], // Very high correlation
        vec![0.038, 0.04, 0.038],
        vec![0.037, 0.038, 0.04],
    ];

    PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .expect("Failed to create correlated portfolio")
}

/// Create a portfolio with singular covariance matrix
fn create_singular_portfolio() -> PortfolioOptimizer {
    let assets = vec!["P".to_string(), "Q".to_string()];
    let returns = vec![0.10, 0.10];
    let covariance = vec![
        vec![0.04, 0.04], // Perfectly correlated (singular)
        vec![0.04, 0.04],
    ];

    PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .expect("Failed to create singular portfolio")
}

// ==================== BASIC PORTFOLIO CREATION TESTS ====================

#[test]
fn test_portfolio_optimizer_creation_valid() {
    let optimizer = create_simple_portfolio();
    assert_eq!(
        optimizer
            .optimize(OptimizationMethod::MinimumVariance)
            .is_ok(),
        true
    );
}

#[test]
fn test_portfolio_optimizer_creation_mismatched_returns() {
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.10]; // Wrong size
    let covariance = vec![vec![0.04, 0.00], vec![0.00, 0.09]];

    let result = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    );

    assert!(result.is_err());
}

#[test]
fn test_portfolio_optimizer_creation_non_square_covariance() {
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.10, 0.15];
    let covariance = vec![
        vec![0.04, 0.00, 0.01], // Extra column
        vec![0.00, 0.09],
    ];

    let result = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    );

    assert!(result.is_err());
}

#[test]
fn test_portfolio_optimizer_creation_mismatched_covariance_size() {
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.10, 0.15];
    let covariance = vec![
        vec![0.04, 0.00, 0.00], // Wrong dimensions
        vec![0.00, 0.09, 0.00],
        vec![0.00, 0.00, 0.05],
    ];

    let result = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    );

    assert!(result.is_err());
}

// ==================== PORTFOLIO METRICS TESTS ====================

#[test]
fn test_portfolio_return_calculation() {
    let optimizer = create_simple_portfolio();
    let weights = vec![0.4, 0.3, 0.3];

    let portfolio_return = optimizer.portfolio_return(&weights);

    // Expected: 0.4*0.10 + 0.3*0.12 + 0.3*0.08 = 0.04 + 0.036 + 0.024 = 0.10
    assert_relative_eq!(portfolio_return, 0.10, epsilon = 1e-6);
}

#[test]
fn test_portfolio_variance_calculation() {
    let optimizer = create_uncorrelated_portfolio();
    let weights = vec![0.5, 0.5];

    let variance = optimizer.portfolio_variance(&weights);

    // Expected: 0.5^2 * 0.04 + 0.5^2 * 0.09 = 0.01 + 0.0225 = 0.0325
    assert_relative_eq!(variance, 0.0325, epsilon = 1e-6);
}

#[test]
fn test_portfolio_volatility_calculation() {
    let optimizer = create_uncorrelated_portfolio();
    let weights = vec![0.5, 0.5];

    let volatility = optimizer.portfolio_volatility(&weights);

    // Expected: sqrt(0.0325) ≈ 0.1803
    assert_relative_eq!(volatility, 0.1803, epsilon = 1e-3);
}

#[test]
fn test_sharpe_ratio_calculation() {
    let optimizer = create_simple_portfolio();
    let weights = vec![0.4, 0.3, 0.3];

    let sharpe = optimizer.sharpe_ratio(&weights);

    // Should be positive for reasonable portfolio
    assert!(sharpe > 0.0);
}

#[test]
fn test_sharpe_ratio_zero_volatility() {
    // Portfolio with zero volatility (all weight in risk-free asset)
    let assets = vec!["RF".to_string()];
    let returns = vec![0.02]; // Risk-free return
    let covariance = vec![vec![0.0]]; // Zero variance

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let sharpe = optimizer.sharpe_ratio(&[1.0]);
    assert_eq!(sharpe, 0.0); // Zero Sharpe for zero excess return
}

// ==================== MEAN-VARIANCE OPTIMIZATION TESTS ====================

#[test]
fn test_mean_variance_optimization_basic() {
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    assert_eq!(result.weights.len(), 3);
    assert!(result.converged);

    // Weights should sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);

    // All weights should be non-negative (long-only)
    for w in &result.weights {
        assert!(*w >= 0.0);
    }
}

#[test]
fn test_mean_variance_optimization_uncorrelated() {
    let optimizer = create_uncorrelated_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Optimization should produce valid weights that sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);

    // Both assets should have non-negative weights
    assert!(result.weights[0] >= 0.0);
    assert!(result.weights[1] >= 0.0);
}

#[test]
fn test_mean_variance_optimization_negative_returns() {
    // Portfolio with negative expected returns (short-only scenario)
    let assets = vec!["DOWN1".to_string(), "DOWN2".to_string()];
    let returns = vec![-0.10, -0.05]; // Negative returns
    let covariance = vec![vec![0.04, 0.01], vec![0.01, 0.09]];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Weights should still sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

// ==================== MINIMUM VARIANCE OPTIMIZATION TESTS ====================

#[test]
fn test_minimum_variance_optimization() {
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MinimumVariance)
        .unwrap();

    assert_eq!(result.weights.len(), 3);
    assert!(result.converged);

    // Weights should sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);

    // Check that this is indeed minimum variance
    let variance = optimizer.portfolio_variance(&result.weights);
    assert!(variance > 0.0);
}

#[test]
fn test_minimum_variance_single_asset() {
    // Single asset portfolio
    let assets = vec!["SOLO".to_string()];
    let returns = vec![0.10];
    let covariance = vec![vec![0.04]];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MinimumVariance)
        .unwrap();

    // Should allocate 100% to the single asset
    assert_relative_eq!(result.weights[0], 1.0, epsilon = 1e-6);
}

#[test]
fn test_minimum_variance_with_singular_matrix() {
    let optimizer = create_singular_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MinimumVariance)
        .unwrap();

    // Should handle singular matrix gracefully (equal weights fallback)
    assert_eq!(result.weights.len(), 2);
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

// ==================== MAXIMUM SHARPE RATIO TESTS ====================

#[test]
fn test_maximum_sharpe_optimization() {
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MaximumSharpe)
        .unwrap();

    assert_eq!(result.weights.len(), 3);

    // Weights should sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);

    // Sharpe ratio should be positive
    assert!(result.sharpe_ratio > 0.0);
}

#[test]
fn test_maximum_sharpe_vs_minimum_variance() {
    let optimizer = create_simple_portfolio();

    let sharpe_result = optimizer
        .optimize(OptimizationMethod::MaximumSharpe)
        .unwrap();
    let minvar_result = optimizer
        .optimize(OptimizationMethod::MinimumVariance)
        .unwrap();

    // Maximum Sharpe should have higher or equal Sharpe ratio
    assert!(sharpe_result.sharpe_ratio >= minvar_result.sharpe_ratio);
}

// ==================== KELLY CRITERION TESTS ====================

#[test]
fn test_kelly_criterion_optimization() {
    let optimizer = create_simple_portfolio();
    let result = optimizer.optimize(OptimizationMethod::Kelly).unwrap();

    assert_eq!(result.weights.len(), 3);

    // Weights should sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_kelly_criterion_growth_optimal() {
    // Kelly criterion should maximize geometric growth
    let assets = vec!["GROWTH".to_string(), "VALUE".to_string()];
    let returns = vec![0.15, 0.08]; // Growth has higher return
    let covariance = vec![vec![0.09, 0.01], vec![0.01, 0.04]]; // Growth has higher vol

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer.optimize(OptimizationMethod::Kelly).unwrap();

    // Should allocate to both assets
    assert!(result.weights[0] > 0.0);
    assert!(result.weights[1] > 0.0);
}

#[test]
fn test_kelly_criterion_with_zero_returns() {
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.0, 0.0]; // Zero expected returns
    let covariance = vec![vec![0.04, 0.00], vec![0.00, 0.09]];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer.optimize(OptimizationMethod::Kelly).unwrap();

    // Should fall back to equal weights
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

// ==================== RISK PARITY TESTS ====================

#[test]
fn test_risk_parity_optimization() {
    let optimizer = create_simple_portfolio();
    let result = optimizer.optimize(OptimizationMethod::RiskParity).unwrap();

    assert_eq!(result.weights.len(), 3);

    // Weights should sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);

    // All weights should be positive
    for w in &result.weights {
        assert!(*w > 0.0);
    }
}

#[test]
fn test_risk_parity_equal_risk_contribution() {
    // Two assets with different volatilities
    let assets = vec!["LOW_VOL".to_string(), "HIGH_VOL".to_string()];
    let returns = vec![0.08, 0.12];
    let covariance = vec![
        vec![0.01, 0.00], // 10% vol
        vec![0.00, 0.09], // 30% vol
    ];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer.optimize(OptimizationMethod::RiskParity).unwrap();

    // Higher volatility asset should have lower weight
    assert!(result.weights[0] > result.weights[1]);
}

#[test]
fn test_risk_parity_volatility_weighted() {
    let optimizer = create_simple_portfolio();
    let result = optimizer.optimize(OptimizationMethod::RiskParity).unwrap();

    // Risk parity should weight inversely to volatility
    // GOOGL (30% vol) should have lower weight than AAPL (20% vol)
    assert!(result.weights[0] > result.weights[1]);
}

// ==================== BLACK-LITTERMAN TESTS ====================

#[test]
fn test_black_litterman_optimization() {
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::BlackLitterman)
        .unwrap();

    assert_eq!(result.weights.len(), 3);

    // Weights should sum to 1.0
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_black_litterman_equilibrium_returns() {
    // Without views, should use equilibrium (market cap) returns
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::BlackLitterman)
        .unwrap();

    // Should converge
    assert!(result.converged);
}

// ==================== CONSTRAINT HANDLING TESTS ====================

#[test]
fn test_long_only_constraint() {
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // All weights should be non-negative (long-only)
    for w in &result.weights {
        assert!(*w >= 0.0, "Weight should be non-negative: {}", w);
    }
}

#[test]
fn test_max_position_constraint() {
    let assets = vec!["A".to_string(), "B".to_string(), "C".to_string()];
    let returns = vec![0.20, 0.10, 0.05]; // A has much higher return
    let covariance = vec![
        vec![0.04, 0.00, 0.00],
        vec![0.00, 0.04, 0.00],
        vec![0.00, 0.00, 0.04],
    ];

    let mut constraints = PortfolioConstraints::default();
    constraints.max_weight = 0.4; // Maximum 40% per asset

    let optimizer =
        PortfolioOptimizer::new(assets, returns, covariance, 0.02, constraints).unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // No weight should exceed 40% (with small tolerance for floating point)
    for w in &result.weights {
        assert!(*w <= 0.401, "Weight {} exceeds max constraint of 0.4", w);
    }
}

#[test]
fn test_minimum_position_constraint() {
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.10, 0.15];
    let covariance = vec![vec![0.04, 0.00], vec![0.00, 0.09]];

    let mut constraints = PortfolioConstraints::default();
    constraints.min_weight = 0.2; // Minimum 20% per asset

    let optimizer =
        PortfolioOptimizer::new(assets, returns, covariance, 0.02, constraints).unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // All weights should be >= 20%
    for w in &result.weights {
        assert!(*w >= 0.2 - 1e-6, "Weight {} below min constraint", w);
    }
}

#[test]
fn test_leverage_constraint() {
    let optimizer = create_simple_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Total weight should equal 1.0 (no leverage)
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

// ==================== EDGE CASE TESTS ====================

#[test]
fn test_empty_portfolio() {
    // Empty portfolio should fail gracefully
    let assets: Vec<String> = vec![];
    let returns: Vec<f64> = vec![];
    let covariance: Vec<Vec<f64>> = vec![];

    let result = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    );

    // Should handle empty portfolio
    assert!(result.is_ok() || result.is_err()); // Either way is acceptable
}

#[test]
fn test_single_asset_portfolio() {
    let assets = vec!["ONLY".to_string()];
    let returns = vec![0.10];
    let covariance = vec![vec![0.04]];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Should allocate 100% to single asset
    assert_relative_eq!(result.weights[0], 1.0, epsilon = 1e-6);
}

#[test]
fn test_highly_correlated_assets() {
    let optimizer = create_correlated_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Should handle multicollinearity gracefully
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_singular_covariance_matrix() {
    let optimizer = create_singular_portfolio();
    let result = optimizer
        .optimize(OptimizationMethod::MaximumSharpe)
        .unwrap();

    // Should fall back to equal weights or similar
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_zero_variance_asset() {
    // Asset with zero variance (risk-free)
    let assets = vec!["RF".to_string(), "RISKY".to_string()];
    let returns = vec![0.02, 0.12];
    let covariance = vec![
        vec![0.0, 0.0], // Risk-free has zero variance
        vec![0.0, 0.09],
    ];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Should handle zero variance gracefully
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_negative_risk_free_rate() {
    // Negative risk-free rate (like Japan, Europe in 2020s)
    let assets = vec!["A".to_string(), "B".to_string()];
    let returns = vec![0.05, 0.08];
    let covariance = vec![vec![0.04, 0.01], vec![0.01, 0.09]];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        -0.005,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MaximumSharpe)
        .unwrap();

    // Should handle negative rates
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
    assert_eq!(result.risk_free_rate, -0.005);
}

// ==================== TRANSACTION COST TESTS ====================

#[test]
fn test_transaction_cost_calculation() {
    let optimizer = create_simple_portfolio();

    let current_weights = vec![0.5, 0.3, 0.2];
    let target_weights = vec![0.4, 0.4, 0.2];

    let cost = optimizer.transaction_costs(&current_weights, &target_weights);

    // Turnover = |0.5-0.4| + |0.3-0.4| + |0.2-0.2| = 0.1 + 0.1 + 0.0 = 0.2
    // Cost = 0.2 * 5 / 10000 = 0.0001
    assert_relative_eq!(cost, 0.0001, epsilon = 1e-8);
}

#[test]
fn test_transaction_cost_zero_turnover() {
    let optimizer = create_simple_portfolio();

    let weights = vec![0.4, 0.3, 0.3];
    let cost = optimizer.transaction_costs(&weights, &weights);

    // No turnover = no cost
    assert_eq!(cost, 0.0);
}

#[test]
fn test_transaction_cost_full_rebalance() {
    let optimizer = create_simple_portfolio();

    let current_weights = vec![1.0, 0.0, 0.0]; // All in first asset
    let target_weights = vec![0.0, 0.0, 1.0]; // All in last asset

    let cost = optimizer.transaction_costs(&current_weights, &target_weights);

    // Full rebalance = 2.0 turnover (sell all of first, buy all of last)
    // Cost = 2.0 * 5 / 10000 = 0.001
    assert_relative_eq!(cost, 0.001, epsilon = 1e-8);
}

#[test]
fn test_transaction_cost_mismatched_lengths() {
    let optimizer = create_simple_portfolio();

    let current_weights = vec![0.5, 0.5];
    let target_weights = vec![0.4, 0.3, 0.3];

    let cost = optimizer.transaction_costs(&current_weights, &target_weights);

    // Should return 0 for mismatched lengths
    assert_eq!(cost, 0.0);
}

// ==================== EFFICIENT FRONTIER TESTS ====================

#[test]
fn test_efficient_frontier_generation() {
    let optimizer = create_simple_portfolio();
    let frontier = optimizer.efficient_frontier(10).unwrap();

    assert_eq!(frontier.len(), 10);

    // All portfolios should have weights summing to 1.0
    for result in &frontier {
        let sum: f64 = result.weights.iter().sum();
        assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
    }
}

#[test]
fn test_efficient_frontier_single_point() {
    let optimizer = create_simple_portfolio();
    let frontier = optimizer.efficient_frontier(1).unwrap();

    assert_eq!(frontier.len(), 1);
}

#[test]
fn test_efficient_frontier_zero_points() {
    let optimizer = create_simple_portfolio();
    let result = optimizer.efficient_frontier(0);

    // Should return error for zero points
    assert!(result.is_err());
}

#[test]
fn test_efficient_frontier_monotonicity() {
    // Frontier points should have monotonically increasing risk-return
    let optimizer = create_simple_portfolio();
    let frontier = optimizer.efficient_frontier(20).unwrap();

    // Check that volatility generally increases with return
    for i in 1..frontier.len() {
        // Allow some tolerance for numerical optimization
        let return_increase = frontier[i].expected_return >= frontier[i - 1].expected_return - 0.01;
        assert!(
            return_increase,
            "Return should be non-decreasing along frontier"
        );
    }
}

// ==================== NUMERICAL STABILITY TESTS ====================

#[test]
fn test_numerical_stability_large_numbers() {
    // Portfolio with large covariance values
    let assets = vec!["BIG1".to_string(), "BIG2".to_string()];
    let returns = vec![0.10, 0.15];
    let covariance = vec![
        vec![1000.0, 100.0], // Large variances
        vec![100.0, 2000.0],
    ];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MinimumVariance)
        .unwrap();

    // Should handle large numbers
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_numerical_stability_small_numbers() {
    // Portfolio with very small covariance values
    let assets = vec!["SMALL1".to_string(), "SMALL2".to_string()];
    let returns = vec![0.001, 0.002];
    let covariance = vec![
        vec![0.0001, 0.00001], // Small variances
        vec![0.00001, 0.0002],
    ];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.0001,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MinimumVariance)
        .unwrap();

    // Should handle small numbers
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

#[test]
fn test_numerical_stability_ill_conditioned_matrix() {
    // Ill-conditioned covariance matrix (high condition number)
    let assets = vec!["ILL1".to_string(), "ILL2".to_string(), "ILL3".to_string()];
    let returns = vec![0.10, 0.11, 0.12];
    let covariance = vec![
        vec![1.0, 0.99, 0.98],
        vec![0.99, 1.0, 0.99],
        vec![0.98, 0.99, 1.0],
    ];

    let optimizer = PortfolioOptimizer::new(
        assets,
        returns,
        covariance,
        0.02,
        PortfolioConstraints::default(),
    )
    .unwrap();

    let result = optimizer
        .optimize(OptimizationMethod::MeanVariance)
        .unwrap();

    // Should handle ill-conditioned matrices
    let sum: f64 = result.weights.iter().sum();
    assert_relative_eq!(sum, 1.0, epsilon = 1e-6);
}

// ==================== CONVEXITY TESTS ====================

#[test]
fn test_portfolio_variance_convexity() {
    // Portfolio variance should be convex in weights
    let optimizer = create_simple_portfolio();

    let w1 = vec![0.5, 0.3, 0.2];
    let w2 = vec![0.3, 0.5, 0.2];
    let lambda = 0.5;

    // Convex combination of weights
    let w_mid: Vec<f64> = w1
        .iter()
        .zip(w2.iter())
        .map(|(a, b)| lambda * a + (1.0 - lambda) * b)
        .collect();

    let var1 = optimizer.portfolio_variance(&w1);
    let var2 = optimizer.portfolio_variance(&w2);
    let var_mid = optimizer.portfolio_variance(&w_mid);

    // Convexity: var(λw1 + (1-λ)w2) <= λvar(w1) + (1-λ)var(w2)
    let upper_bound = lambda * var1 + (1.0 - lambda) * var2;
    assert!(
        var_mid <= upper_bound + 1e-6,
        "Portfolio variance should be convex"
    );
}

#[test]
fn test_portfolio_return_linearity() {
    // Portfolio return should be linear in weights
    let optimizer = create_simple_portfolio();

    let w1 = vec![0.5, 0.3, 0.2];
    let w2 = vec![0.3, 0.5, 0.2];
    let lambda = 0.5;

    let w_mid: Vec<f64> = w1
        .iter()
        .zip(w2.iter())
        .map(|(a, b)| lambda * a + (1.0 - lambda) * b)
        .collect();

    let ret1 = optimizer.portfolio_return(&w1);
    let ret2 = optimizer.portfolio_return(&w2);
    let ret_mid = optimizer.portfolio_return(&w_mid);

    // Linearity: ret(λw1 + (1-λ)w2) = λret(w1) + (1-λ)ret(w2)
    let expected = lambda * ret1 + (1.0 - lambda) * ret2;
    assert_relative_eq!(ret_mid, expected, epsilon = 1e-6);
}

// ==================== REBALANCING FREQUENCY TESTS ====================

#[test]
fn test_rebalancing_frequency_high_turnover() {
    // High transaction costs should favor lower turnover
    let optimizer = create_simple_portfolio();

    let current_weights = vec![0.6, 0.2, 0.2];
    let target_weights = vec![0.2, 0.4, 0.4]; // High turnover

    let cost = optimizer.transaction_costs(&current_weights, &target_weights);

    // High turnover = high cost (default is 5 bps, turnover is 0.8)
    // Cost = 0.8 * 5 / 10000 = 0.0004
    assert!(cost > 0.0003); // More than 3 basis points (realistic threshold)
}

#[test]
fn test_rebalancing_frequency_low_turnover() {
    let optimizer = create_simple_portfolio();

    let current_weights = vec![0.35, 0.35, 0.30];
    let target_weights = vec![0.33, 0.33, 0.34]; // Low turnover

    let cost = optimizer.transaction_costs(&current_weights, &target_weights);

    // Low turnover = low cost
    assert!(cost < 0.0002); // Less than 2 basis points
}