quantwave-core 0.1.6

use crate::indicators::metadata::{IndicatorMetadata, ParamDef};
use crate::traits::Next;
use crate::indicators::bandpass::BandPass;
use std::collections::VecDeque;

/// Voss Predictive Filter
///
/// Based on John Ehlers' "A Peek Into The Future".
/// Uses a two-pole bandpass filter followed by a Voss predictor to achieve
/// negative group delay for band-limited signals.
#[derive(Debug, Clone)]
pub struct VossPredictor {
    bandpass: BandPass,
    order: usize,
    voss_history: VecDeque<f64>,
}

impl VossPredictor {
    pub fn new(period: usize, predict: usize) -> Self {
        let order = 3 * predict;
        Self {
            bandpass: BandPass::new(period, 0.25), // Bandwidth default 0.25 as per paper
            order,
            voss_history: VecDeque::with_capacity(order + 1),
        }
    }
}

impl Default for VossPredictor {
    fn default() -> Self {
        Self::new(20, 3)
    }
}

impl Next<f64> for VossPredictor {
    type Output = (f64, f64); // (Filt, Voss)

    fn next(&mut self, input: f64) -> Self::Output {
        let filt = self.bandpass.next(input);
        
        let mut sum_c = 0.0;
        if self.order > 0 {
            for count in 0..self.order {
                let idx = self.order - count;
                // voss_history[0] is Voss[1] (value 1 bar ago)
                // voss_history[idx - 1] is Voss[idx]
                let val = if idx <= self.voss_history.len() {
                    self.voss_history[idx - 1]
                } else {
                    0.0
                };
                sum_c += ((count + 1) as f64 / self.order as f64) * val;
            }
        }

        let voss = ((3.0 + self.order as f64) / 2.0) * filt - sum_c;
        
        self.voss_history.push_front(voss);
        if self.voss_history.len() > self.order {
            self.voss_history.pop_back();
        }

        (filt, voss)
    }
}

pub const VOSS_PREDICTOR_METADATA: IndicatorMetadata = IndicatorMetadata {
    name: "VossPredictor",
    description: "A predictive filter with negative group delay for band-limited signals.",
    params: &[
        ParamDef { name: "period", default: "20", description: "Center period of the BandPass filter" },
        ParamDef { name: "predict", default: "3", description: "Number of bars of prediction" },
    ],
    formula_source: "https://github.com/lavs9/quantwave/blob/main/references/Ehlers%20Papers/A%20PEEK%20INTO%20THE%20FUTURE.pdf",
    formula_latex: r#"
\[
Filt = \text{BandPass}(Price, Period, 0.25)
\]
\[
Order = 3 \cdot Predict
\]
\[
SumC = \sum_{n=0}^{Order-1} \frac{n+1}{Order} Voss_{t-(Order-n)}
\]
\[
Voss = \frac{3 + Order}{2} Filt - SumC
\]
"#,
    gold_standard_file: "voss_predictor.json",
    category: "Ehlers DSP",
};

#[cfg(test)]
mod tests {
    use super::*;
    use crate::traits::Next;
    use crate::test_utils::{load_gold_standard_tuple, assert_indicator_parity_tuple};
    use proptest::prelude::*;

    #[test]
    fn test_voss_gold_standard() {
        let case = load_gold_standard_tuple("voss_predictor");
        let vp = VossPredictor::new(20, 3);
        assert_indicator_parity_tuple(vp, &case.input, &case.expected);
    }

    #[test]
    fn test_voss_basic() {
        let mut vp = VossPredictor::default();
        let inputs = vec![10.0, 11.0, 12.0, 13.0, 14.0, 15.0];
        for input in inputs {
            let (filt, voss) = vp.next(input);
            assert!(!filt.is_nan());
            assert!(!voss.is_nan());
        }
    }

    proptest! {
        #[test]
        fn test_voss_parity(
            inputs in prop::collection::vec(1.0..100.0, 60..120),
        ) {
            let period = 20;
            let predict = 3;
            let mut vp = VossPredictor::new(period, predict);
            let streaming_results: Vec<(f64, f64)> = inputs.iter().map(|&x| vp.next(x)).collect();
            
            // Batch implementation
            let mut batch_results = Vec::with_capacity(inputs.len());
            let mut bp = BandPass::new(period, 0.25);
            let order = 3 * predict;
            let mut v_hist = VecDeque::new();
            
            for &input in &inputs {
                let filt = bp.next(input);
                let mut sum_c = 0.0;
                for count in 0..order {
                    let idx = order - count;
                    let val = if idx <= v_hist.len() {
                        v_hist[idx - 1]
                    } else {
                        0.0
                    };
                    sum_c += ((count + 1) as f64 / order as f64) * val;
                }
                
                let voss = ((3.0 + order as f64) / 2.0) * filt - sum_c;
                v_hist.push_front(voss);
                if v_hist.len() > order {
                    v_hist.pop_back();
                }
                batch_results.push((filt, voss));
            }
            
            for (s, b) in streaming_results.iter().zip(batch_results.iter()) {
                approx::assert_relative_eq!(s.0, b.0, epsilon = 1e-10);
                approx::assert_relative_eq!(s.1, b.1, epsilon = 1e-10);
            }
        }
    }
}