Realized Volatility Jump

Realized volatility jumps are detected through time-series decomposition of high-frequency returns into continuous and jump components. In insider-trading surveillance and quant scoring, a realized volatility jump preceding or coinciding with Form 4 filings, Rule 10b5-1 plan adoptions, or blackout-window breaches can flag suspicious timing. The jump magnitude, duration, and volume profile are critical diagnostics: benign microstructure noise produces small, rapidly-decaying jumps, whereas information-driven jumps exhibit persistence and correlation with subsequent price moves.

Quantitatively, realized volatility jumps are isolated using bipower variation, threshold-based identification, or Bayesian change-point methods. On an insider-scoring platform, the anomaly is weighted by proximity to disclosure events, executive rank (PDMR status), sector momentum regime, and cross-sectional factor exposure. A realized volatility jump in a low-liquidity or thinly-traded security carries higher suspicion weight than one in a high-volume liquid name, controlling for market-microstructure-noise and tick-size-pilot effects.