Jump Detection

Jump Detection is the definitive clean build of jump detection for everyday charts: each bar's log return is divided by a robust estimate of ordinary volatility and plotted as a Standardized Return in sigma units against dashed lines at ±3σ. Bars clearing it print Jump Markers and optional price-chart labels, and a dashboard splits recent variance into continuous and jump-driven parts.

How to Trade the Jump Detection?

• Reading beyond ±3σ: the bar repriced beyond what its trailing benchmark calls ordinary - volatility estimates, band tags and stops spanning it deserve separate treatment.
• Gap Jump alert: the open itself repriced beyond the threshold from the prior close, so price skipped the levels between - where a stop could not have filled.
• Jump Share of Variance climbing: the dashboard shows variance arriving in lurches rather than steady churn.

An isolated dislocation stands out sharply while a sustained stretch of large bars lifts the benchmark and stops being flagged. Four alerts cover up, down, either-direction and gap jumps - useful pause triggers for the volatility toolkit.

Jump Detection Settings

• Volatility Lookback (default 20): trailing bars behind the benchmark, ending on the prior bar so the tested bar never inflates its own yardstick.
• Volatility Estimator (default Bipower Variation): the robust scale for standardizing; Median Absolute Deviation and Standard Deviation are the alternatives.
• Jump Threshold (σ) (default 3): sigma multiple a return must exceed; raise toward 4-5 on noisy intraday data.
• Mark Jumps on Chart (default on): labels jumps on the price chart with magnitudes.
• Show Dashboard (default on): the variance-decomposition panel.
• Gradient Fill (default on): style toggle for the zero-to-value fill.

Frequently Asked Questions

How is a jump different from range expansion?

A jump is a single bar out of line with its surroundings; expansion is a persistent run of larger bars, which lifts the robust benchmark and stops flagging. ATR Expansion & Contraction is the dedicated tool for that regime shift itself.

Which volatility estimator should I choose?

Bipower Variation is the default because products of absolute adjacent returns barely react to one outsized bar. Median Absolute Deviation behaves similarly; Standard Deviation shows the failure mode: one jump inflates it for a full window, muting the detector after an event.

Does the indicator predict jumps?

No - it identifies and measures them once the bar prints. The value is in treatment: cleaning volatility estimates and separating band tags that landed across a jump bar.