Skip to main content

Stability Features

Time-series stability diagnostics. These FeatureTypes measure how a signal behaves over time — autocorrelation decay, half-life, turnover, and term structure.

FeatureCategoryDescription
autocorrelationstatisticsAutocorrelation of any input signal at specified lag — measures signal persis...
consecutive_above_thresholdstatisticsDuration of consecutive observations above a threshold — measures regime pers...
liquidity_recovery_half_lifeorder_flowHalf-life of liquidity recovery after stress — time for depth to recover from...
order_flow_autocorrelationorder_flowOrder flow autocorrelation: corr(ofi, prev(ofi, 1), window=20)
range_expansion_ratemomentumRate of price range expansion — measures whether volatility is expanding or c...
rolling_volatility_of_volatilityvolatilityRolling volatility of volatility: rolling_std(vol_of_vol) — measures stabilit...
signal_half_lifestatisticsHalf-life of signal persistence — time for signal to drop below threshold
signal_turnoverstatisticsSignal turnover: fraction of signal direction flips per window — low turnover...
volatility_term_structure_slopevolatilitySlope of volatility term structure — measures how vol decays with time scale