CDRR ranks a fixed 20-asset universe by current derivatives stress. It combines volatility, leverage movement, funding crowding, liquidations, and venue concentration into one cross-sectional score while keeping market direction in a separate state engine.
The score is designed to answer a narrow question: where is derivatives stress currently concentrated within the tracked universe? It is therefore a ranking and monitoring system, not a return-prediction model.
Each component is normalised to the interval [0, 1]. Production v1 deliberately uses simple equal weights while longer forward-stress history is collected.
How elevated and how rapidly expanding realised volatility is.
Available mean of the asset's 24H historical volatility percentile and its current cross-sectional volatility-expansion percentile.
Higher values indicate an asset is already volatile relative to its own recent history, the current universe, or both.
The magnitude of current leverage build-up or contraction.
Cross-sectional percentile of the absolute 1H common-universe open-interest change.
The component measures leverage movement magnitude. Direction is handled separately by the state engine.
How unusually one-sided perpetual positioning appears through funding.
Confirmed funding-tail percentile using cross-sectional and prior historical calibration once the history gate is ready.
The factor is based on crowding magnitude; the sign of funding remains descriptive context rather than score direction.
Current forced-position stress relative to both time and the cross-section.
Conservative confirmation of the temporal 1H liquidation percentile with the cross-sectional percentile of 1H liquidations divided by clean open interest.
A high value requires liquidation activity to look abnormal through more than one normalisation.
How concentrated tracked derivatives open interest is across venues.
Cross-sectional percentile of the clean open-interest Herfindahl-Hirschman Index.
More concentrated venue exposure receives a higher concentration component.
Conditional variance uses a finite power-law memory over the latest 250 five-minute squared returns:
Student-t residuals allow heavier tails than a Gaussian specification.
Twelve five-minute conditional variance steps are aggregated to form the next-one-hour forecast. The terminal displays the result as annualised volatility and also reports the implied one-hour sigma.
A tuned exponentially weighted variance model runs beside the primary forecast.
The spread is a model-disagreement diagnostic, not an additional production risk factor.
OI change is calculated only on contracts that are valid both now and at the comparison horizon.
Contract identity uses symbol, market identifier, and exchange identifier.
Production leverage stress uses the cross-sectional percentile of the absolute one-hour change.
A provider adding or dropping contracts can move raw aggregate OI even when the actual positions in comparable contracts have not changed. Common-universe matching removes that coverage contamination.
Historical percentiles use a midrank empirical CDF so ties are handled symmetrically.
The current observation is not allowed to calibrate its own historical percentile.
Funding crowding is calibrated against both the current cross-section and the asset's prior history. Once the history gate is ready, the production signal keeps the more conservative confirmation.
Funding rates are not annualised in the terminal.
Production liquidation stress combines abnormal activity through time with current liquidations relative to open interest across assets.
Price is rising while common-universe open interest is expanding.
Price is falling while common-universe open interest is expanding.
Open interest is contracting materially as positions are reduced.
Forced liquidations are elevated and dominated by long positions.
Forced liquidations are elevated and dominated by short positions.
Liquidation stress and open-interest contraction coincide.
Stress is elevated while leverage is building into positive price action. The system does not convert that combination into a bearish prediction.
Stress is elevated while price weakness and leverage dynamics point to a more defensive or forced-risk regime.
The market remains stressed, but leverage is being removed rather than added. That is a different regime from fresh leveraged build-up.
Provider market observations older than 15 minutes are rejected from the current derivatives snapshot.
The contract base symbol must match the tracked asset, and duplicate symbol / market pairs are reduced to the newest observation.
CMC market-pair outliers are rejected. Price or volume exclusion does not automatically invalidate otherwise usable OI, funding, or basis fields.
Tracked market-pair OI is checked against exchange-wide derivatives OI when that reference is available. Missing exchange-wide OI is allowed but marked unverified.
The live fifth factor currently uses the cross-sectional OI concentration percentile. This keeps the production score stable while the newer market-structure channels accumulate longitudinal evidence.
Basis magnitude and cross-venue basis dispersion are now historically calibrated, but the candidate remains outside the production score until sufficient forward-stress and longitudinal history supports promotion.
Ranks are only relative to the 20 tracked assets. They are not a statement about the entire crypto market.
Several calibration and forward-validation datasets are still accumulating. Production weights are intentionally conservative while that history grows.
High-frequency forward-validation snapshots overlap heavily, so a large snapshot count does not imply the same number of independent statistical observations.
CDRR estimates stress magnitude and market regime. It does not produce expected returns, price targets, or trade instructions.