CDRR
● LOADING
PRIMARY VOL MODEL?
POWER-LAW
STUDENT-t · 1H CONDITIONAL
MEDIAN MODEL SPREAD?
—
PL VS EWMA
PL ABOVE EWMA?
0/20
CURRENT SNAPSHOT
HIGH VOL NAMES?
0/20
24H VOL ≥ 80P
RISK MODEL READY?
0/20
PROVISIONAL v1 · EQUAL WEIGHT
MODEL MONITOR
VOLATILITY FORECASTING · FACTOR DECOMPOSITION · RESEARCH STATUS
LIVE MODEL OUTPUTS · STATIC RESEARCH VALIDATION LABELLED SEPARATELY
01PRIMARY
POWER-LAW VOL

Student-t conditional variance model with power-law memory. Twelve 5-minute variance forecasts are aggregated into the next 1-hour conditional volatility estimate.

02SECONDARY
EWMA VOL

Tuned exponentially weighted variance model used as a secondary benchmark. The live model spread measures disagreement against the primary power-law forecast.

03PRODUCTION
CDRR v1

Provisional equal-weight stress score: volatility, leverage/open interest, funding crowding, liquidations, and concentration. Direction is kept separate in the state engine.

04RESEARCH ONLY
STRUCTURE v2

Candidate structure channel: 50% OI concentration, 25% confirmed basis magnitude, 25% confirmed basis dispersion. It is not used in the production score.

OFFLINE VOLATILITY VALIDATION
WALK-FORWARD RESEARCH · NOT A LIVE PERFORMANCE METRIC
LOWER MEAN QLIKE
17 / 20
POWER-LAW VS EWMA
SIGNIFICANT PL WINS
5
HAC-ROBUST TEST
SIGNIFICANT EWMA WINS
0
HAC-ROBUST TEST
FORECAST HORIZON
1H
12 × 5-MIN STEPS
MODEL DETAIL
CURRENT VOLATILITY OUTPUTS AND CDRR FACTOR LOADS
SELECT AN ASSET FROM THE TABLE
0 / 0 ASSETS
LIVE VOLATILITY MODEL MONITOR
REALISED VOL · CONDITIONAL FORECASTS · MODEL DISAGREEMENT
● LOADING
—
LOADING MODEL OUTPUTS...
PL = POWER-LAW STUDENT-t MODELEWMA = SECONDARY BENCHMARKFORECAST FIGURES ARE ANNUALISED VOLATILITY1H σ = IMPLIED ONE-HOUR STANDARD DEVIATIONMODEL SPREAD = (PL − EWMA) / EWMA
CDRR FACTOR MATRIX
PROVISIONAL v1 · EQUAL-WEIGHT 5-FACTOR STRESS MAGNITUDE
RANKASSETSCOREVOLOIFUNDLIQCONCFUND SOURCESTATE
PRODUCTION FACTOR D = OI CONCENTRATION ONLY. BASIS LEVEL AND BASIS DISPERSION REMAIN RESEARCH-ONLY UNTIL LONGITUDINAL VALIDATION SUPPORTS PROMOTION.