The `Next<T>` Pattern

Next<Input> is QuantWave's universal indicator trait. One implementation powers streaming state machines, Polars batch columns, and gold-standard tests.

Short answer

Implement Next<f64> (or Next<&Bar> for rich signals), call .next(input) per bar, and batch Polars paths will match if they share the same core logic.

Trait definition

pub trait Next<Input> {
    type Output;
    fn next(&mut self, input: Input) -> Self::Output;
}

Source: quantwave_core::traits::Next — see docs.rs.

Streaming RSI example

use quantwave_core::indicators::RSI;
use quantwave_core::traits::Next;

fn main() {
    let mut rsi = RSI::new(14);
    let prices = [44.0, 44.5, 43.8, 44.2, 45.0];

    for price in prices {
        let value = rsi.next(price);
        println!("RSI: {value:?}");
    }
}

Warmup bars return None until the lookback is satisfied — same semantics as Polars .ta().rsi(period=14).

Why it matters

Layer	How `Next<T>` is used
Streaming	Stateful struct, one `next()` per tick
Polars batch	Expression plugin walks the series with identical math
Tests	`proptest` asserts `batch == streaming.collect()`
Backtest (advanced)	`Next<&Bar, Output = StrategySignal>` drives streaming simulation

SmoothingAlgorithm — swappable MA inside composite indicators (SuperTrend, Keltner, …)
IndicatorConfig — builds a fresh Next state machine from parameters

Steel-thread reference

SuperTrend is the canonical end-to-end indicator: core Next impl, Polars .ta(), gold-standard JSON, and documentation.

The Next<T> Pattern