foxhunt/ml/tests/volume_bars_test.rs

//! TDD Tests for Volume Bar Sampling
//!
//! Tests volume-based bar formation (bars emitted when volume threshold reached)
//!
//! ## Test Coverage
//! 1. Basic volume accumulation and bar formation
//! 2. OHLCV calculation correctness
//! 3. Adaptive threshold (EWMA of recent bar volumes)
//! 4. Edge cases: single large trade, zero volume periods
//! 5. Performance: <50μs per bar formation
//! 6. Consistency: volume per bar should match threshold

use chrono::Utc;
use ml::features::alternative_bars::{OHLCVBar as AltBar, VolumeBarSampler};
use std::time::Instant;

#[test]
fn test_volume_bar_basic_formation() {
    // Fixed threshold: 1000 contracts
    let mut sampler = VolumeBarSampler::new(1000.0, false);

    // Trade 1: 300 contracts at $4500 (11:00:00)
    let ts1 = Utc::now();
    let bar1 = sampler.update(4500.0, 300.0, ts1);
    assert!(bar1.is_none(), "Bar should not form yet (300/1000)");

    // Trade 2: 400 contracts at $4505 (11:00:05)
    let ts2 = ts1 + chrono::Duration::seconds(5);
    let bar2 = sampler.update(4505.0, 400.0, ts2);
    assert!(bar2.is_none(), "Bar should not form yet (700/1000)");

    // Trade 3: 350 contracts at $4510 (11:00:10) -> Exceeds 1000
    let ts3 = ts2 + chrono::Duration::seconds(5);
    let bar3 = sampler.update(4510.0, 350.0, ts3);
    assert!(bar3.is_some(), "Bar should form (1050 >= 1000)");

    let bar = bar3.unwrap();
    assert_eq!(bar.open, 4500.0, "Open should be first trade price");
    assert_eq!(bar.high, 4510.0, "High should be max price");
    assert_eq!(bar.low, 4500.0, "Low should be min price");
    assert_eq!(bar.close, 4510.0, "Close should be last trade price");
    assert_eq!(bar.volume, 1050.0, "Volume should be sum of trades");
    assert_eq!(bar.timestamp, ts1, "Timestamp should be bar start");

    println!("✅ Basic volume bar formation validated");
}

#[test]
fn test_volume_bar_ohlcv_correctness() {
    let mut sampler = VolumeBarSampler::new(500.0, false);

    // Scenario: 5 trades forming 2 complete bars
    let ts_base = Utc::now();
    let trades = vec![
        // Bar 1 (600 volume)
        (4500.0, 100.0, ts_base),                                // Open=4500
        (4510.0, 150.0, ts_base + chrono::Duration::seconds(1)), // High=4510
        (4495.0, 200.0, ts_base + chrono::Duration::seconds(2)), // Low=4495
        (4505.0, 150.0, ts_base + chrono::Duration::seconds(3)), // Close=4505
        // Bar 2 (550 volume)
        (4508.0, 100.0, ts_base + chrono::Duration::seconds(4)), // Open=4508
        (4520.0, 250.0, ts_base + chrono::Duration::seconds(5)), // High=4520
        (4507.0, 100.0, ts_base + chrono::Duration::seconds(6)), // Low=4507
        (4515.0, 100.0, ts_base + chrono::Duration::seconds(7)), // Close=4515
    ];

    let mut bars = Vec::new();
    for (price, volume, ts) in trades {
        if let Some(bar) = sampler.update(price, volume, ts) {
            bars.push(bar);
        }
    }

    assert_eq!(bars.len(), 2, "Should form 2 complete bars");

    // Bar 1 validation
    assert_eq!(bars[0].open, 4500.0, "Bar 1: Wrong open");
    assert_eq!(bars[0].high, 4510.0, "Bar 1: Wrong high");
    assert_eq!(bars[0].low, 4495.0, "Bar 1: Wrong low");
    assert_eq!(bars[0].close, 4505.0, "Bar 1: Wrong close");
    assert_eq!(bars[0].volume, 600.0, "Bar 1: Wrong volume");

    // Bar 2 validation
    assert_eq!(bars[1].open, 4508.0, "Bar 2: Wrong open");
    assert_eq!(bars[1].high, 4520.0, "Bar 2: Wrong high");
    assert_eq!(bars[1].low, 4507.0, "Bar 2: Wrong low");
    assert_eq!(bars[1].close, 4515.0, "Bar 2: Wrong close");
    assert_eq!(bars[1].volume, 550.0, "Bar 2: Wrong volume");

    println!("✅ OHLCV calculation correctness validated");
}

#[test]
fn test_volume_bar_adaptive_threshold() {
    // Adaptive threshold: uses EWMA of recent bar volumes
    let mut sampler = VolumeBarSampler::new(1000.0, true); // Enable adaptive

    // First bar: 1200 volume (exceeds initial threshold)
    let ts_base = Utc::now();
    let bar1 = sampler.update(4500.0, 600.0, ts_base);
    assert!(bar1.is_none());
    let bar1 = sampler.update(4505.0, 600.0, ts_base + chrono::Duration::seconds(1));
    assert!(bar1.is_some());
    assert_eq!(bar1.as_ref().unwrap().volume, 1200.0);

    // Second bar: threshold should adapt towards 1200
    // EWMA(α=0.2): new_threshold = 0.2 * 1200 + 0.8 * 1000 = 1040
    let bar2 = sampler.update(4510.0, 520.0, ts_base + chrono::Duration::seconds(2));
    assert!(
        bar2.is_none(),
        "Should not form bar yet with adaptive threshold"
    );
    let bar2 = sampler.update(4515.0, 530.0, ts_base + chrono::Duration::seconds(3));
    assert!(bar2.is_some(), "Should form bar (1050 > ~1040)");

    println!("✅ Adaptive threshold EWMA validated");
}

#[test]
fn test_volume_bar_single_large_trade() {
    // Edge case: Single trade exceeds threshold
    let mut sampler = VolumeBarSampler::new(1000.0, false);

    let ts = Utc::now();
    let bar = sampler.update(4500.0, 5000.0, ts);

    assert!(bar.is_some(), "Should form bar immediately");
    let bar = bar.unwrap();

    assert_eq!(bar.open, 4500.0);
    assert_eq!(bar.high, 4500.0);
    assert_eq!(bar.low, 4500.0);
    assert_eq!(bar.close, 4500.0);
    assert_eq!(bar.volume, 5000.0, "Should capture full large trade");

    println!("✅ Single large trade edge case validated");
}

#[test]
fn test_volume_bar_zero_volume_handling() {
    // Edge case: Zero volume trades (should be ignored or handled gracefully)
    let mut sampler = VolumeBarSampler::new(1000.0, false);

    let ts = Utc::now();

    // Zero volume trade
    let bar1 = sampler.update(4500.0, 0.0, ts);
    assert!(bar1.is_none(), "Zero volume should not contribute");

    // Normal trades after zero volume
    let bar2 = sampler.update(4505.0, 500.0, ts + chrono::Duration::seconds(1));
    assert!(bar2.is_none());
    let bar3 = sampler.update(4510.0, 500.0, ts + chrono::Duration::seconds(2));
    assert!(bar3.is_some());

    let bar = bar3.unwrap();
    assert_eq!(bar.volume, 1000.0, "Should only count non-zero volumes");
    assert_eq!(bar.open, 4505.0, "Should ignore zero-volume price in OHLC");

    println!("✅ Zero volume handling validated");
}

#[test]
fn test_volume_bar_performance() {
    // Performance: <50μs per bar formation
    let mut sampler = VolumeBarSampler::new(10000.0, false);

    let ts_base = Utc::now();
    let num_trades = 10000;

    let start = Instant::now();
    let mut bars_formed = 0;

    for i in 0..num_trades {
        let price = 4500.0 + (i as f64 % 100.0) * 0.1;
        let volume = 5.0; // Small increments to test many updates
        let ts = ts_base + chrono::Duration::milliseconds(i);

        if sampler.update(price, volume, ts).is_some() {
            bars_formed += 1;
        }
    }

    let elapsed = start.elapsed();
    let avg_per_update = elapsed.as_nanos() as f64 / num_trades as f64;
    let avg_per_bar = if bars_formed > 0 {
        elapsed.as_nanos() as f64 / bars_formed as f64
    } else {
        0.0
    };

    println!("✅ Performance test completed:");
    println!("   - Total trades: {}", num_trades);
    println!("   - Bars formed: {}", bars_formed);
    println!("   - Avg per update: {:.2}ns", avg_per_update);
    println!(
        "   - Avg per bar: {:.2}ns ({:.2}μs)",
        avg_per_bar,
        avg_per_bar / 1000.0
    );

    // Target: <50μs per bar formation
    assert!(
        avg_per_bar < 50_000.0,
        "Performance target missed: {:.2}μs > 50μs",
        avg_per_bar / 1000.0
    );
}

#[test]
fn test_volume_bar_consistency() {
    // Consistency: volume per bar should match threshold (±1 trade)
    let threshold = 1000.0;
    let mut sampler = VolumeBarSampler::new(threshold, false);

    let ts_base = Utc::now();
    let mut bars = Vec::new();

    // Simulate 5000 trades, each 50 contracts
    for i in 0..5000 {
        let price = 4500.0 + (i as f64).sin() * 10.0;
        let volume = 50.0;
        let ts = ts_base + chrono::Duration::milliseconds(i);

        if let Some(bar) = sampler.update(price, volume, ts) {
            bars.push(bar);
        }
    }

    // Each bar should have volume close to threshold
    for (i, bar) in bars.iter().enumerate() {
        assert!(
            bar.volume >= threshold && bar.volume <= threshold + 50.0,
            "Bar {} volume {} outside expected range [{}, {}]",
            i,
            bar.volume,
            threshold,
            threshold + 50.0
        );
    }

    println!(
        "✅ Volume consistency validated: {} bars formed",
        bars.len()
    );
    println!(
        "   - Volume range: {:.2} - {:.2}",
        bars.iter().map(|b| b.volume).fold(f64::INFINITY, f64::min),
        bars.iter()
            .map(|b| b.volume)
            .fold(f64::NEG_INFINITY, f64::max)
    );
}

#[test]
fn test_volume_bar_time_interval_variance() {
    // Volume bars should have varying time intervals (high activity = faster bars)
    let mut sampler = VolumeBarSampler::new(1000.0, false);

    let ts_base = Utc::now();
    let mut bars = Vec::new();

    // Simulate varying activity: first 10 bars fast, next 10 bars slow
    let mut ts = ts_base;

    // Fast activity: 100 contracts per second (10s per bar)
    for _ in 0..10 {
        for _ in 0..10 {
            ts = ts + chrono::Duration::seconds(1);
            if let Some(bar) = sampler.update(4500.0, 100.0, ts) {
                bars.push((bar, ts));
                break;
            }
        }
    }

    // Slow activity: 10 contracts per second (100s per bar)
    for _ in 0..10 {
        for _ in 0..100 {
            ts = ts + chrono::Duration::seconds(1);
            if let Some(bar) = sampler.update(4500.0, 10.0, ts) {
                bars.push((bar, ts));
                break;
            }
        }
    }

    // Validate time intervals vary
    assert_eq!(bars.len(), 20, "Should form 20 bars");

    let fast_intervals: Vec<_> = bars
        .iter()
        .take(10)
        .zip(bars.iter().skip(1).take(9))
        .map(|((_, ts1), (_, ts2))| (*ts2 - *ts1).num_seconds())
        .collect();

    let slow_intervals: Vec<_> = bars
        .iter()
        .skip(10)
        .take(9)
        .zip(bars.iter().skip(11).take(9))
        .map(|((_, ts1), (_, ts2))| (*ts2 - *ts1).num_seconds())
        .collect();

    let avg_fast = fast_intervals.iter().sum::<i64>() as f64 / fast_intervals.len() as f64;
    let avg_slow = slow_intervals.iter().sum::<i64>() as f64 / slow_intervals.len() as f64;

    println!("✅ Time interval variance validated:");
    println!("   - Fast activity: avg {:.2}s per bar", avg_fast);
    println!("   - Slow activity: avg {:.2}s per bar", avg_slow);

    assert!(
        avg_slow > avg_fast * 5.0,
        "Slow bars should take significantly longer than fast bars"
    );
}

#[test]
fn test_volume_bar_multiple_bar_sequence() {
    // Integration test: Process 1000 trades, validate all bars
    let mut sampler = VolumeBarSampler::new(500.0, false);

    let ts_base = Utc::now();
    let mut bars = Vec::new();

    for i in 0..1000 {
        let price = 4500.0 + (i as f64 * 0.1).sin() * 50.0;
        let volume = 10.0 + (i as f64 * 0.05).cos() * 5.0; // Varying volume
        let ts = ts_base + chrono::Duration::milliseconds(i * 100);

        if let Some(bar) = sampler.update(price, volume, ts) {
            bars.push(bar);
        }
    }

    // Should form ~100 bars (1000 trades * ~10-15 volume / 500 threshold)
    assert!(
        bars.len() >= 20 && bars.len() <= 35,
        "Expected 20-35 bars, got {}",
        bars.len()
    );

    // All bars should be valid
    for (i, bar) in bars.iter().enumerate() {
        assert!(bar.open > 0.0, "Bar {} has invalid open", i);
        assert!(bar.high >= bar.low, "Bar {} has high < low", i);
        assert!(bar.high >= bar.open, "Bar {} has high < open", i);
        assert!(bar.high >= bar.close, "Bar {} has high < close", i);
        assert!(bar.low <= bar.open, "Bar {} has low > open", i);
        assert!(bar.low <= bar.close, "Bar {} has low > close", i);
        assert!(bar.volume > 0.0, "Bar {} has zero volume", i);
    }

    println!("✅ Multiple bar sequence validated: {} bars", bars.len());
}