use crate::error::FinError;
use crate::signals::{BarInput, Signal, SignalValue};
use rust_decimal::prelude::ToPrimitive;
use rust_decimal::Decimal;
use std::collections::VecDeque;
pub struct LinearRegressionSlope {
name: String,
period: usize,
closes: VecDeque<Decimal>,
}
impl LinearRegressionSlope {
pub fn new(name: impl Into<String>, period: usize) -> Result<Self, FinError> {
if period < 2 {
return Err(FinError::InvalidPeriod(period));
}
Ok(Self {
name: name.into(),
period,
closes: VecDeque::with_capacity(period),
})
}
}
impl Signal for LinearRegressionSlope {
fn name(&self) -> &str { &self.name }
fn period(&self) -> usize { self.period }
fn is_ready(&self) -> bool { self.closes.len() >= self.period }
fn update(&mut self, bar: &BarInput) -> Result<SignalValue, FinError> {
self.closes.push_back(bar.close);
if self.closes.len() > self.period {
self.closes.pop_front();
}
if self.closes.len() < self.period {
return Ok(SignalValue::Unavailable);
}
let n = self.period as f64;
let t_mean = (n - 1.0) / 2.0;
let mut sum_tx = 0.0_f64;
let mut sum_tt = 0.0_f64;
for (i, c) in self.closes.iter().enumerate() {
let t = i as f64;
let x = match c.to_f64() {
Some(v) => v,
None => return Ok(SignalValue::Unavailable),
};
let dt = t - t_mean;
sum_tx += dt * x;
sum_tt += dt * dt;
}
if sum_tt == 0.0 {
return Ok(SignalValue::Unavailable);
}
let slope = sum_tx / sum_tt;
Decimal::try_from(slope)
.map(SignalValue::Scalar)
.or(Ok(SignalValue::Unavailable))
}
fn reset(&mut self) {
self.closes.clear();
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::ohlcv::OhlcvBar;
use crate::signals::Signal;
use crate::types::{NanoTimestamp, Price, Quantity, Symbol};
use rust_decimal_macros::dec;
fn bar(c: &str) -> OhlcvBar {
let p = Price::new(c.parse().unwrap()).unwrap();
OhlcvBar {
symbol: Symbol::new("X").unwrap(),
open: p, high: p, low: p, close: p,
volume: Quantity::zero(),
ts_open: NanoTimestamp::new(0),
ts_close: NanoTimestamp::new(1),
tick_count: 1,
}
}
#[test]
fn test_lrs_invalid_period() {
assert!(LinearRegressionSlope::new("lrs", 0).is_err());
assert!(LinearRegressionSlope::new("lrs", 1).is_err());
}
#[test]
fn test_lrs_unavailable_before_period() {
let mut s = LinearRegressionSlope::new("lrs", 3).unwrap();
assert_eq!(s.update_bar(&bar("100")).unwrap(), SignalValue::Unavailable);
assert_eq!(s.update_bar(&bar("101")).unwrap(), SignalValue::Unavailable);
assert!(!s.is_ready());
}
#[test]
fn test_lrs_perfect_uptrend_gives_positive_slope() {
let mut s = LinearRegressionSlope::new("lrs", 3).unwrap();
s.update_bar(&bar("100")).unwrap();
s.update_bar(&bar("101")).unwrap();
if let SignalValue::Scalar(v) = s.update_bar(&bar("102")).unwrap() {
assert!((v - dec!(1)).abs() < dec!(0.001), "perfect uptrend slope should be 1: {v}");
} else {
panic!("expected Scalar");
}
}
#[test]
fn test_lrs_perfect_downtrend_gives_negative_slope() {
let mut s = LinearRegressionSlope::new("lrs", 3).unwrap();
s.update_bar(&bar("102")).unwrap();
s.update_bar(&bar("101")).unwrap();
if let SignalValue::Scalar(v) = s.update_bar(&bar("100")).unwrap() {
assert!(v < dec!(0), "downtrend should give negative slope: {v}");
} else {
panic!("expected Scalar");
}
}
#[test]
fn test_lrs_flat_gives_near_zero() {
let mut s = LinearRegressionSlope::new("lrs", 4).unwrap();
for _ in 0..4 { s.update_bar(&bar("100")).unwrap(); }
if let SignalValue::Scalar(v) = s.update_bar(&bar("100")).unwrap() {
assert!(v.abs() < dec!(0.001), "flat prices should give ~zero slope: {v}");
} else {
panic!("expected Scalar");
}
}
#[test]
fn test_lrs_reset() {
let mut s = LinearRegressionSlope::new("lrs", 3).unwrap();
for p in &["100","101","102"] { s.update_bar(&bar(p)).unwrap(); }
assert!(s.is_ready());
s.reset();
assert!(!s.is_ready());
}
}