Overnight Gap

Close-to-open return across the session boundary: open / previous_close - 1, detected automatically at each local day boundary and held for the session.

Quick reference

Item	Value
Family	Seasonality & Session
Input type	Candle (uses open, close, timestamp)
Output type	f64 (simple return)
Output range	unbounded (typically small, e.g. ±0.05)
Default parameters	utc_offset_minutes = 0 (UTC)
Warmup period	2
Interpretation	Overnight repricing; gap-fill / continuation setups

Formula

At each new local day:
  gap = open_first_bar / close_last_bar_of_prior_day - 1
held constant until the next day boundary

The first session yields no gap. See crates/wickra-core/src/indicators/overnight_gap.rs.

Parameters

Name	Type	Default	Constraint	Source	Description
utc_offset_minutes	i32	0	none	overnight_gap.rs:34	Shifts the instant before deriving the day boundary.

OvernightGap::new is infallible.

Inputs / Outputs

const _: fn(&mut wickra::OvernightGap, wickra::Candle) -> Option<f64> =
    <wickra::OvernightGap as wickra::Indicator>::update;

• Python. update((o,h,l,c,v,ts)) → float | None; batch(...) → 1-D ndarray (NaN warmup).
• Node. update(...) → number | null; batch(...) → number[].
• WASM. update(...) → number | undefined.

Warmup

warmup_period() == 2. The first session has no prior close, so it returns None; from the first boundary crossing onward a value is held for the session.

Edge cases

• First session has no gap (test first_session_has_no_gap).
• Gap computed at the day boundary and held (test computes_gap_at_day_boundary).
• Negative (gap-down) values are supported (test negative_gap_down).
• Zero prior close yields a 0.0 gap rather than a non-finite value (test zero_prev_close_yields_zero_gap).
• Reset clears the prior close and gap (test reset_clears_state).

Examples

Rust

use wickra::{Candle, Indicator, OvernightGap};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let day = 24 * 3_600_000;
    let mut gap = OvernightGap::new(0);
    gap.update(Candle::new(99.0, 101.0, 98.0, 100.0, 1.0, 0)?);            // day 1 closes 100
    let g = gap.update(Candle::new(105.0, 106.0, 104.0, 105.5, 1.0, day)?).unwrap();
    println!("{g}"); // 0.05
    Ok(())
}

Python

import wickra as ta

DAY = 24 * 3_600_000
gap = ta.OvernightGap(0)
assert gap.update((99.0, 101.0, 98.0, 100.0, 1.0, 0)) is None
print(gap.update((105.0, 106.0, 104.0, 105.5, 1.0, DAY)))  # 0.05

Node

const wickra = require('wickra');
const DAY = 24 * 3_600_000;
const gap = new wickra.OvernightGap(0);
gap.update(99, 101, 98, 100, 1, 0);
console.log(gap.update(105, 106, 104, 105.5, 1, DAY)); // 0.05

Streaming

gap = ta.OvernightGap(-300)  # US Eastern session open
for o, h, l, c, v, ts in candle_feed:
    g = gap.update((o, h, l, c, v, ts))
    if g is not None and abs(g) > 0.02:
        pass  # large overnight gap — fade (gap-fill) or follow (continuation)

Interpretation

1. Gap-fill vs continuation. Small gaps tend to fill intraday; large news-driven gaps more often continue. The gap magnitude is the key signal.
2. Overnight risk. The gap quantifies the portion of return earned while the position was held but unmonitored.

Common pitfalls

• Session definition. For equities the "overnight" is the RTH close→open; set utc_offset_minutes so the day boundary matches the session, not UTC.
• Continuous markets. On 24/7 instruments the "gap" is whatever happens across the chosen midnight boundary, which may be economically meaningless.

See also

• OvernightIntradayReturn — splits the full return into overnight + intraday legs.
• Indicators-Overview — full taxonomy.