foxhunt/ml/examples/wave_comparison_simple.rs

//! Simplified Wave Comparison Backtest (Feature Quality Assessment)
//!
//! This backtest compares Wave C (65 features) vs Wave D (225 features)
//! using a simple momentum strategy to evaluate feature quality improvements.
//!
//! Strategy: Buy when momentum > threshold, sell when momentum < -threshold
//! This isolates the impact of feature engineering from model complexity.
//!
//! Usage:
//!   cargo run -p ml --example wave_comparison_simple --release

use anyhow::Result;
use chrono::{DateTime, Utc};
use common::ml_strategy::MLFeatureExtractor;
use data::providers::databento::dbn_parser::{DbnParser, ProcessedMessage};
use num_traits::ToPrimitive;
use std::path::PathBuf;

/// Performance metrics for backtest
#[derive(Debug, Clone)]
struct PerformanceMetrics {
    total_trades: usize,
    winning_trades: usize,
    win_rate: f64,
    total_pnl: f64,
    total_return: f64,
    sharpe_ratio: f64,
    max_drawdown: f64,
    calmar_ratio: f64,
    profit_factor: f64,
}

/// Trade record
#[derive(Debug, Clone)]
struct Trade {
    entry_time: DateTime<Utc>,
    exit_time: DateTime<Utc>,
    entry_price: f64,
    exit_price: f64,
    side: TradeSide,
    pnl: f64,
}

#[derive(Debug, Clone, Copy)]
enum TradeSide {
    Long,
    Short,
}

/// Market data bar
#[derive(Debug, Clone)]
struct MarketBar {
    timestamp: DateTime<Utc>,
    open: f64,
    high: f64,
    low: f64,
    close: f64,
    volume: f64,
}

/// Load market data from DBN file
fn load_market_data(dbn_path: &PathBuf) -> Result<Vec<MarketBar>> {
    println!("📖 Loading market data from: {}", dbn_path.display());

    let parser = DbnParser::new()
        .map_err(|e| anyhow::anyhow!("Failed to create DBN parser: {}", e))?;

    let dbn_bytes = std::fs::read(dbn_path)?;
    let messages = parser
        .parse_batch(&dbn_bytes)
        .map_err(|e| anyhow::anyhow!("Failed to parse DBN file: {}", e))?;

    let mut bars = Vec::new();

    for msg in messages {
        if let ProcessedMessage::Ohlcv {
            symbol: _,
            timestamp,
            open,
            high,
            low,
            close,
            volume,
        } = msg
        {
            let ts_secs = (timestamp.as_nanos() / 1_000_000_000) as i64;
            bars.push(MarketBar {
                timestamp: DateTime::from_timestamp(ts_secs, 0).unwrap_or_else(|| Utc::now()),
                open: open.to_f64(),
                high: high.to_f64(),
                low: low.to_f64(),
                close: close.to_f64(),
                volume: volume.to_f64().unwrap_or(0.0),
            });
        }
    }

    bars.sort_by_key(|bar| bar.timestamp);
    println!("✅ Loaded {} bars", bars.len());

    Ok(bars)
}

/// Calculate maximum drawdown from equity curve
fn calculate_max_drawdown(equity_curve: &[f64]) -> f64 {
    if equity_curve.is_empty() {
        return 0.0;
    }

    let mut max_drawdown = 0.0;
    let mut peak = equity_curve[0];

    for &equity in equity_curve {
        if equity > peak {
            peak = equity;
        }
        let drawdown = (peak - equity) / peak;
        if drawdown > max_drawdown {
            max_drawdown = drawdown;
        }
    }

    max_drawdown
}

/// Extract momentum signal from feature vector
/// Uses features 5-10 (technical indicators like RSI, MACD) for signal generation
fn extract_momentum_signal(features: &[f64]) -> f64 {
    if features.len() < 10 {
        return 0.0;
    }

    // Combine multiple feature signals
    // Features 5-10 typically include RSI, MACD, momentum indicators
    let signal1 = features.get(5).cloned().unwrap_or(0.0);
    let signal2 = features.get(6).cloned().unwrap_or(0.0);
    let signal3 = features.get(7).cloned().unwrap_or(0.0);
    let signal4 = features.get(8).cloned().unwrap_or(0.0);
    let signal5 = features.get(9).cloned().unwrap_or(0.0);

    // Average and normalize
    let avg = (signal1 + signal2 + signal3 + signal4 + signal5) / 5.0;
    avg.clamp(-1.0, 1.0)
}

/// Run momentum-based backtest
fn run_backtest(
    feature_extractor: &mut MLFeatureExtractor,
    market_data: &[MarketBar],
    initial_capital: f64,
    wave_name: &str,
) -> Result<PerformanceMetrics> {
    println!("\n🔄 Running {} backtest...", wave_name);

    let mut trades = Vec::new();
    let mut position: Option<(TradeSide, f64, DateTime<Utc>, f64)> = None;
    let mut equity_curve = vec![initial_capital];
    let mut current_capital = initial_capital;

    // Simple momentum strategy parameters
    let signal_threshold = 0.15; // Lower threshold for more trades
    let holding_periods = 20;    // Hold for ~20 bars

    let mut bars_in_position = 0;

    for bar in market_data.iter() {
        // Extract features
        let features = feature_extractor.extract_features(
            bar.close,
            bar.volume,
            bar.timestamp,
        );

        // Get momentum signal from features
        let signal = extract_momentum_signal(&features);

        // Trading logic
        if position.is_none() && signal.abs() > signal_threshold {
            // Enter position
            let side = if signal > 0.0 {
                TradeSide::Long
            } else {
                TradeSide::Short
            };
            let size = (current_capital * 0.1) / bar.close; // 10% of capital
            position = Some((side, size, bar.timestamp, bar.close));
            bars_in_position = 0;
        } else if let Some((side, size, entry_time, entry_price)) = position {
            bars_in_position += 1;

            // Exit logic: signal reversal or holding period exceeded
            let should_exit = match side {
                TradeSide::Long => signal < -0.1 || bars_in_position >= holding_periods,
                TradeSide::Short => signal > 0.1 || bars_in_position >= holding_periods,
            };

            if should_exit {
                let pnl = match side {
                    TradeSide::Long => size * (bar.close - entry_price),
                    TradeSide::Short => size * (entry_price - bar.close),
                };

                current_capital += pnl;
                equity_curve.push(current_capital);

                trades.push(Trade {
                    entry_time,
                    exit_time: bar.timestamp,
                    entry_price,
                    exit_price: bar.close,
                    side,
                    pnl,
                });

                position = None;
                bars_in_position = 0;
            }
        }
    }

    // Close any open position at the end
    if let Some((side, size, entry_time, entry_price)) = position {
        let last_bar = &market_data[market_data.len() - 1];
        let pnl = match side {
            TradeSide::Long => size * (last_bar.close - entry_price),
            TradeSide::Short => size * (entry_price - last_bar.close),
        };
        current_capital += pnl;
        equity_curve.push(current_capital);

        trades.push(Trade {
            entry_time,
            exit_time: last_bar.timestamp,
            entry_price,
            exit_price: last_bar.close,
            side,
            pnl,
        });
    }

    // Calculate metrics
    let total_trades = trades.len();
    let winning_trades = trades.iter().filter(|t| t.pnl > 0.0).count();
    let win_rate = if total_trades > 0 {
        (winning_trades as f64 / total_trades as f64) * 100.0
    } else {
        0.0
    };
    let total_pnl: f64 = trades.iter().map(|t| t.pnl).sum();
    let total_return = (current_capital - initial_capital) / initial_capital * 100.0;

    // Sharpe ratio (annualized)
    let returns: Vec<f64> = trades.iter().map(|t| t.pnl / initial_capital).collect();
    let sharpe_ratio = if !returns.is_empty() {
        let mean_return = returns.iter().sum::<f64>() / returns.len() as f64;
        let variance = returns
            .iter()
            .map(|r| (r - mean_return).powi(2))
            .sum::<f64>()
            / returns.len() as f64;
        let std_dev = variance.sqrt();
        if std_dev > 0.0 {
            (mean_return / std_dev) * (252.0_f64).sqrt()
        } else {
            0.0
        }
    } else {
        0.0
    };

    // Max drawdown
    let max_drawdown = calculate_max_drawdown(&equity_curve) * 100.0;

    // Calmar ratio
    let calmar_ratio = if max_drawdown > 0.0 {
        total_return / max_drawdown
    } else {
        0.0
    };

    // Profit factor
    let gross_profit: f64 = trades.iter().filter(|t| t.pnl > 0.0).map(|t| t.pnl).sum();
    let gross_loss: f64 = trades
        .iter()
        .filter(|t| t.pnl < 0.0)
        .map(|t| t.pnl.abs())
        .sum();
    let profit_factor = if gross_loss > 0.0 {
        gross_profit / gross_loss
    } else if gross_profit > 0.0 {
        f64::INFINITY
    } else {
        0.0
    };

    Ok(PerformanceMetrics {
        total_trades,
        winning_trades,
        win_rate,
        total_pnl,
        total_return,
        sharpe_ratio,
        max_drawdown,
        calmar_ratio,
        profit_factor,
    })
}

fn print_metrics(metrics: &PerformanceMetrics, wave_name: &str) {
    println!("\n{}", "=".repeat(70));
    println!("📈 {} RESULTS", wave_name.to_uppercase());
    println!("{}", "=".repeat(70));
    println!("\n💰 Performance Metrics:");
    println!("   Total Trades:      {}", metrics.total_trades);
    println!("   Winning Trades:    {}", metrics.winning_trades);
    println!("   Win Rate:          {:.2}%", metrics.win_rate);
    println!("   Total PnL:         ${:.2}", metrics.total_pnl);
    println!("   Total Return:      {:.2}%", metrics.total_return);
    println!("   Sharpe Ratio:      {:.2}", metrics.sharpe_ratio);
    println!("   Max Drawdown:      {:.2}%", metrics.max_drawdown);
    println!("   Calmar Ratio:      {:.2}", metrics.calmar_ratio);
    println!("   Profit Factor:     {:.2}", metrics.profit_factor);
}

fn print_comparison(wave_c: &PerformanceMetrics, wave_d: &PerformanceMetrics) {
    println!("\n{}", "=".repeat(70));
    println!("📊 WAVE C vs WAVE D COMPARISON");
    println!("{}", "=".repeat(70));

    let sharpe_improvement = ((wave_d.sharpe_ratio - wave_c.sharpe_ratio) / wave_c.sharpe_ratio.max(0.01)) * 100.0;
    let win_rate_improvement = wave_d.win_rate - wave_c.win_rate;
    let drawdown_improvement = ((wave_c.max_drawdown - wave_d.max_drawdown) / wave_c.max_drawdown.max(0.01)) * 100.0;
    let return_improvement = wave_d.total_return - wave_c.total_return;

    println!("\n🎯 Key Improvements:");
    println!("   Sharpe Ratio:      {:.2} → {:.2} ({:+.1}%)",
        wave_c.sharpe_ratio, wave_d.sharpe_ratio, sharpe_improvement);
    println!("   Win Rate:          {:.2}% → {:.2}% ({:+.1}pp)",
        wave_c.win_rate, wave_d.win_rate, win_rate_improvement);
    println!("   Max Drawdown:      {:.2}% → {:.2}% ({:+.1}%)",
        wave_c.max_drawdown, wave_d.max_drawdown, drawdown_improvement);
    println!("   Total Return:      {:.2}% → {:.2}% ({:+.2}pp)",
        wave_c.total_return, wave_d.total_return, return_improvement);

    println!("\n✅ Target Validation:");
    println!("   Sharpe ≥ 2.0:      {} (actual: {:.2})",
        if wave_d.sharpe_ratio >= 2.0 { "✅ PASS" } else { "❌ FAIL" },
        wave_d.sharpe_ratio);
    println!("   Win Rate ≥ 60%:    {} (actual: {:.2}%)",
        if wave_d.win_rate >= 60.0 { "✅ PASS" } else { "❌ FAIL" },
        wave_d.win_rate);
    println!("   Drawdown ≤ 15%:    {} (actual: {:.2}%)",
        if wave_d.max_drawdown <= 15.0 { "✅ PASS" } else { "❌ FAIL" },
        wave_d.max_drawdown);

    let all_targets_met = wave_d.sharpe_ratio >= 2.0
        && wave_d.win_rate >= 60.0
        && wave_d.max_drawdown <= 15.0;

    println!("\n{}", if all_targets_met {
        "🎉 ALL TARGETS MET - PRODUCTION READY!"
    } else {
        "⚠️  Some targets not met - further optimization needed"
    });
}

fn main() -> Result<()> {
    println!("\n{}", "=".repeat(70));
    println!("🚀 WAVE COMPARISON BACKTEST (Simplified Feature Quality Assessment)");
    println!("{}\n", "=".repeat(70));

    let data_path = PathBuf::from("/home/jgrusewski/Work/foxhunt/test_data/real/databento/ES.FUT_ohlcv-1m_2024-01-02.dbn");
    let initial_capital = 100000.0;

    // Load market data once
    let market_data = load_market_data(&data_path)?;

    println!("📊 Backtest Configuration:");
    println!("   Symbol:           ES.FUT");
    println!("   Bars:             {}", market_data.len());
    println!("   Initial Capital:  ${:.2}", initial_capital);
    println!("   Strategy:         Simple Momentum (Feature Quality Test)");

    // Run Wave C backtest (65 features)
    let mut wave_c_extractor = MLFeatureExtractor::new_wave_c(20);
    let wave_c_metrics = run_backtest(
        &mut wave_c_extractor,
        &market_data,
        initial_capital,
        "Wave C (65 features)",
    )?;
    print_metrics(&wave_c_metrics, "Wave C (65 features)");

    // Run Wave D backtest (225 features)
    let mut wave_d_extractor = MLFeatureExtractor::new_wave_d(20);
    let wave_d_metrics = run_backtest(
        &mut wave_d_extractor,
        &market_data,
        initial_capital,
        "Wave D (225 features)",
    )?;
    print_metrics(&wave_d_metrics, "Wave D (225 features)");

    // Print comparison
    print_comparison(&wave_c_metrics, &wave_d_metrics);

    println!("\n{}", "=".repeat(70));

    Ok(())
}