CROSS-ASSET VALIDATION
CIF v0.21.3
Frozen-basket MOS replication across liquid altcoins. Same v0.8 MOS formula, no auto-tuning.
Basket
liquid-alt-v1-2026-08-07
liquid-alt-v1-2026-08-07
DIAGNOSTIC, NOT PRODUCTION VALIDATION. The basket is frozen before results are observed, but it is still survivorship-biased. Pooled rows share dates and market regimes, so date-matched cross-sectional and per-asset consistency are more important than one pooled correlation.
FROZEN BASKET
External replication sample
BNBSOLXRPADADOGELINKAVAXLTC
Replay points1736
Mature 30d1616
Coverage ≥70 points1736
Coverage ≥70 mature1616
Build status: success · assets succeeded 8/8 · score engine deterministic-preview-v0.8.
PRIMARY CROSS-ASSET READOUT
Coverage ≥70%
Pooled MOS → 30d excess BTC-0.095N=1616
Pooled BTC tertile spread-2.30 pp
Median per-asset ρ-0.1828 assets
Positive asset share12.5%
DATE-MATCHED CROSS-SECTION
Does higher MOS rank better assets on the same date?
Median date Spearman-0.095
Mean date Spearman-0.105
Positive-date share38.6%
Median Top2−Bottom2 spread-2.01 pp
This is the preferred MOS test in v0.10 because all compared assets share the same checkpoint date and subsequent market regime.
PER-ASSET CONSISTENCY
Does the sign replicate across coins?
AssetN 30dCoverage ≥7030d excess BTC ρBTC tertile spread30d drawdown ρ
BNB2022170.106-1.55-0.138
SOL202217-0.197-5.70-0.043
XRP202217-0.170-5.17-0.189
ADA202217-0.194-7.02-0.215
DOGE202217-0.0410.32-0.226
LINK202217-0.260-9.26-0.283
AVAX202217-0.257-9.45-0.176
LTC202217-0.055-2.67-0.260
COMPONENT REPLICATION
Do HEI component signs generalize?
ComponentPooled NPooled ρMedian asset ρPositive assetsAssets tested
1D Trend1616-0.038-0.11225.0%8
4H / 1H Structure1616-0.046-0.09625.0%8
Volume & Participation16160.0640.078100.0%8
Relative Strength1616-0.229-0.2720.0%8
1M / 1W Structure1616-0.087-0.14225.0%8
Range Positioning1616-0.101-0.21712.5%8
A component that looked weak on HEI should only be considered for future redesign if the same sign replicates across several independent assets and later survives a frozen out-of-sample test.
BTC MARKET REGIME
Does MOS behave differently above/below BTC EMA200?
BTC ≥ EMA200856 mature 30d points
MOS → excess BTC ρ-0.121
BTC tertile spread-2.12 pp
MOS → drawdown ρ-0.114
BTC < EMA200760 mature 30d points
MOS → excess BTC ρ-0.158
BTC tertile spread-2.59 pp
MOS → drawdown ρ-0.134
ANTI-OVERFIT / INTERPRETATION
What v0.10 is allowed to conclude
- Basket was frozen before cross-asset results were observed.
- Basket is survivorship-biased: it uses assets that are liquid/listed in 2026.
- BTC and ETH are benchmarks and are not target assets.
- Cross-asset MOS evidence does not validate AQS or Risk.
- Pooled point statistics are not treated as independent observations because dates and assets share market regimes.
- No production gate, MOS weight, normalization curve or formula is changed by this release.
- Preferred evidence is cross-sectional date-matched behavior plus consistency across individual assets.
- If a redesign is proposed later, it must be frozen before testing on a different/forward sample.