← Validation Lab

Decomposition ↗Candidates ↗

ALPHA SIGNAL RESEARCH LAB

CIF v0.21.3

Test transparent primary factors independently across horizons and axes before any future MOS v2 is allowed to exist.

Factor screen
2020-08 → 2022-08

NO COMBINED SCORE. Eight factor hypotheses, directions and gates are frozen before this screen. A passing factor is only screened in for a new independent confirmation sample.

FROZEN FACTOR SCREEN

Independent earlier time block

success · complete

BNBXRPADADOGELINKLTCTRXBCHETCXLM

Replay points1050

Mature 7d1050

Mature 30d1050

Mature 90d1050

Assets succeeded 10/10. Primary horizon is 30d; 7d and 90d must confirm the sign.

PRIMARY FACTOR LIBRARY

Eight pre-registered hypotheses

8 factors

30D Momentummomentum_30d · momentum

Trailing 30-day close-to-close return.

direction: higher_is_better

30D Relative Momentum vs BTCrelative_momentum_btc_30d · relative_momentum

Asset trailing 30-day return minus BTC trailing 30-day return.

direction: higher_is_better

EMA200 Trend Distancetrend_ema200 · trend

Percent distance of current close above/below trailing EMA200.

direction: higher_is_better

90D Range Positionbreakout_90d · breakout

Current close position inside the trailing 90-day high-low range, 0..100.

direction: higher_is_better

Vol-Adjusted 30D Momentumvol_adjusted_momentum_30d · risk_adjusted_momentum

Trailing 30-day return divided by annualized trailing 30-day realized volatility.

direction: higher_is_better

Volume Expansion 7v30volume_expansion_7v30 · participation

Average quote volume over the last 7 days divided by average quote volume over the preceding 30 days.

direction: higher_is_better

Low Volatility 30Dlow_volatility_30d · low_volatility

Negative annualized trailing 30-day realized volatility; higher means lower volatility.

direction: higher_is_better

7D Mean Reversionmean_reversion_7d · mean_reversion

Negative trailing 7-day return; higher means more oversold over the prior week.

direction: higher_is_better

MULTI-HORIZON SCREEN

Cross-sectional vs time-series evidence

ρ: 7d / 30d / 90d

FactorXS ρTS ρXS +datesTS +assetsStatus

30D Momentummomentum_30d-0.067 / -0.164 / -0.0790.070 / 0.065 / -0.00737.1%70.0%NO

30D Relative Momentum vs BTCrelative_momentum_btc_30d-0.067 / -0.164 / -0.079-0.112 / -0.273 / -0.26037.1%0.0%NO

EMA200 Trend Distancetrend_ema200-0.067 / -0.127 / -0.0670.009 / 0.081 / 0.13542.9%70.0%TS

90D Range Positionbreakout_90d-0.042 / -0.115 / -0.0060.010 / -0.016 / -0.03440.0%50.0%NO

Vol-Adjusted 30D Momentumvol_adjusted_momentum_30d-0.055 / -0.055 / 0.0300.063 / 0.044 / 0.02342.9%60.0%NO

Volume Expansion 7v30volume_expansion_7v30-0.030 / -0.067 / -0.0180.055 / -0.015 / -0.06342.9%40.0%NO

Low Volatility 30Dlow_volatility_30d0.103 / 0.079 / 0.2000.058 / -0.003 / -0.12860.0%50.0%NO

7D Mean Reversionmean_reversion_7d0.067 / -0.055 / -0.006-0.010 / -0.093 / -0.01344.8%10.0%NO

PRIMARY 30D GATES

What actually passed?

7d + 90d sign confirmation required

30D MomentumUNSUPPORTED

XS 30d ρ-0.164spread -5.16 pp

XS gateNO

TS 30d ρ0.06570.0% positive assets

TS gateNO

30D Relative Momentum vs BTCUNSUPPORTED

XS 30d ρ-0.164spread -5.16 pp

XS gateNO

TS 30d ρ-0.2730.0% positive assets

TS gateNO

EMA200 Trend DistanceTIME_SERIES_SUPPORTED

XS 30d ρ-0.127spread -2.72 pp

XS gateNO

TS 30d ρ0.08170.0% positive assets

TS gatePASS

90D Range PositionUNSUPPORTED

XS 30d ρ-0.115spread -0.40 pp

XS gateNO

TS 30d ρ-0.01650.0% positive assets

TS gateNO

Vol-Adjusted 30D MomentumUNSUPPORTED

XS 30d ρ-0.055spread 0.95 pp

XS gateNO

TS 30d ρ0.04460.0% positive assets

TS gateNO

Volume Expansion 7v30UNSUPPORTED

XS 30d ρ-0.067spread -1.38 pp

XS gateNO

TS 30d ρ-0.01540.0% positive assets

TS gateNO

Low Volatility 30DUNSUPPORTED

XS 30d ρ0.079spread 6.71 pp

XS gateNO

TS 30d ρ-0.00350.0% positive assets

TS gateNO

7D Mean ReversionUNSUPPORTED

XS 30d ρ-0.055spread -2.98 pp

XS gateNO

TS 30d ρ-0.09310.0% positive assets

TS gateNO

ORTHOGONALITY / REDUNDANCY

Are factors measuring the same thing?

|ρ| ≥ 0.80 flagged

Factor pairs28

Redundant pairs4

Correlation clusters5

Automatic pruningNO

30D Momentum↔ 30D Relative Momentum vs BTC

same-date ρ1.000

within-asset ρ0.528

30D Momentum↔ Vol-Adjusted 30D Momentum

same-date ρ0.939

within-asset ρ0.973

30D Relative Momentum vs BTC↔ Vol-Adjusted 30D Momentum

same-date ρ0.939

within-asset ρ0.510

EMA200 Trend Distance↔ 90D Range Position

same-date ρ0.709

within-asset ρ0.846

Clusters: 30D Momentum ↔ 30D Relative Momentum vs BTC ↔ Vol-Adjusted 30D Momentum · 90D Range Position ↔ EMA200 Trend Distance · Volume Expansion 7v30 · Low Volatility 30D · 7D Mean Reversion

SCREEN VERDICT

FACTORS_SCREENED_IN_FOR_CONFIRMATION

1/8 screened in

EMA200 Trend DistanceTIME_SERIES_SUPPORTED

Eligible only for a new independent confirmation sample. It is not a production factor.

RESEARCH FIREWALL

What v0.13 is allowed to conclude

The 2020-2022 factor screen is consumed after this first result and cannot be reused as confirmation.
Eight primary factor definitions and their directions were frozen before observing screen results.
30d is the primary horizon; 7d and 90d must confirm the sign for a factor to pass an axis gate.
Cross-sectional tests use subsequent excess vs BTC; time-series tests use subsequent raw asset returns.
No grid search, factor coefficient fitting, MOS weight changes, or combined alpha score is performed in v0.13.
A screened-in factor is only eligible for a new independent confirmation sample.
Redundancy is descriptive; correlated factors are not automatically combined or discarded.
No production gate, AQS formula, Risk formula, or current MOS v0.8 formula changes in v0.13.