What causes "`calculate_from_returns` {IMPL_ERR} `{}`" in nautechsystems/nautilus_trader?

Implement `calculate_from_returns` for the statistic type.

`calculate_from_returns` {IMPL_ERR} `{}`

Error message

`calculate_from_returns` {IMPL_ERR} `{}`

What it means

`PortfolioAnalyzer`/statistic trait's default `calculate_from_returns` panics because the concrete statistic did not override it. Statistics must implement the calculation method for the input type the analyzer will feed them; calling the unimplemented default is a programming error.

Source

Thrown at crates/analysis/src/statistic.rs:48

///
/// The analyzer calls `calculate_from_returns`, `calculate_from_realized_pnls`, and
/// `calculate_from_positions` on every registered statistic, and their defaults panic, so an
/// implementation must override all three and return `None` for a category it does not support.
/// `calculate_from_returns_with_benchmark` defaults to `None` and is optional.
#[allow(unused_variables)]
pub trait PortfolioStatistic: Debug {
    type Item;

    /// Returns the name of this statistic for display and identification purposes.
    fn name(&self) -> String;

    /// Calculates the statistic from time-indexed returns data.
    ///
    /// # Panics
    ///
    /// Panics if this method is not implemented for the specific statistic.
    fn calculate_from_returns(&self, returns: &Returns) -> Option<Self::Item> {
        panic!("`calculate_from_returns` {IMPL_ERR} `{}`", self.name());
    }

    /// Calculates the statistic from realized profit and loss values.
    ///
    /// # Panics
    ///
    /// Panics if this method is not implemented for the specific statistic.
    fn calculate_from_realized_pnls(&self, realized_pnls: &[f64]) -> Option<Self::Item> {
        panic!(
            "`calculate_from_realized_pnls` {IMPL_ERR} `{}`",
            self.name()
        );
    }

    /// Calculates the statistic from position data.
    ///
    /// # Panics
    ///

View on GitHub (pinned to 18893faf8b)

Solutions

Implement `calculate_from_returns` for the statistic type.
Or compute that statistic from the input type it supports (positions or realized PnLs) instead.
If the statistic genuinely cannot be returns-based, guard the analysis path to skip it for returns input.

Example fix

// before
impl PerformanceStatistic for MyStat {
    fn calculate_from_positions(&self, _: &[Position]) -> Option<f64> { Some(0.0) }
}
analyzer.calculate_from_returns(&returns); // panic
// after
impl PerformanceStatistic for MyStat {
    fn calculate_from_returns(&self, returns: &Returns) -> Option<f64> {
        Some(my_stat_from_returns(returns))
    }
    fn calculate_from_positions(&self, _: &[Position]) -> Option<f64> { Some(0.0) }
}

Defensive patterns

Strategy: type-guard

Validate before calling

// pick statistics that support returns input
let stats: Vec<Box<dyn PerformanceStatistic>> = vec![Box::new(SharpeRatio::new(...))];
assert!(stats.iter().all(|s| s.supports_returns()), "statistic does not support returns input");

Type guard

fn is_returns_compatible(stat: &dyn PerformanceStatistic) -> bool {
    // a statistic supports returns if it overrides calculate_from_returns
    stat.calculate_from_returns(&Returns::default()).is_some() || stat.is_returns_based()
}

Prevention

Implement all trait methods (returning None where unsupported) in custom statistics
Register each statistic only into analysis paths matching its implemented methods
Document which input type each custom statistic supports

When it happens

Trigger: Registering a statistic with the portfolio analyzer and computing performance from `Returns` while that statistic only implements `calculate_from_positions` or `calculate_from_realized_pnls`.

Common situations: Custom `PerformanceStatistic` implementations that only override one calculation method; using a position-based statistic (e.g. certain expectancy variants) in a returns-based analysis path.

Related errors

AI-assisted analysis of nautechsystems/nautilus_trader@18893faf8b (2026-09-08). Data as JSON: /api/errors/e338666a1fadf2f9. Report an issue: GitHub.